Ph.D., Tohoku University

Computer simulation

Image processing

Neuroinformatics

Computational neuroscience

Human Informatics

人間情報学・計算論的神経科学・ニューロインフォマティクス・画像処理アルゴリズム・計算機シミュレーショ

包括脳ネットワーク

多重解像度

V1野

初期視覚

画像再構成

filling in

長距離水平結合

固視微動

コントラスト感度

視覚モデル

微分幾何

拡散現象

分数階微分

V4細胞

Life sciences, Neuroscience - general

Informatics, Perceptual information processing

Informatics, Biological, health, and medical informatics

Informatics, Intelligent informatics

Informatics, Sensitivity (kansei) informatics

Informatics, Soft computing

2010 - 2011
The University of Electro-Communications, 准教授

01 Apr. 2009
電気通信大学大学院・情報システム学研究科, 准教授

01 Apr. 2007 - 31 Mar. 2009
理化学研究所・脳科学総合研究センター, 研究員

2008 - 2009
RIKEN, 研究員

2009
電気通信大学 大学院・情報システム学研究科, 准教授

2007 - 2008
RIKEN, 研究員

01 Apr. 2004 - 31 Mar. 2007
東北福祉大学・総合福祉学部・情報福祉学科, 講師（専任）

2005 - 2007
Tohoku Fukushi University, 講師

2004 - 2006
Tohoku Fukushi University, Faculty of Comprehensive Welfare, 講師

01 Oct. 2001 - 01 Apr. 2004
東北大学大学院工学研究科, 助手

2002 - 2004
東北大学 工学(系)研究科(研究院), 助手

2002 - 2003
東北大学 大学院・工学研究科, 助手

01 Apr. 2000 - 30 Sep. 2001
日本学術振興会, 特別研究員（PD）

Apr. 1997 - Mar. 2000
Tohoku University, Graduate School, Division of Engineering, Electrical Engineering, Communication Engineering, Electronic Engineering, and Information Engineering

Mar. 1991
岩手県立黒沢尻北高等学校, 普通科

Apr. 2013
委員長, Visiome Platform, The OECD Working Group on Neuroinfonnatics の方針に基づくプロジェクト研究の一つ．

Jan. 2010
理事, 日本神経回路学会, Society

Mar. 2024
情報処理学会
Simple Color-Correction using a Mirror and a Virtual Color Chart
Student Encouragement Award of IPSJ National Convention, Yuki Kojima;Shunji SATOH

Dec. 2023
映像情報メディア学会
両眼視野闘争の性質を利用した2色型色覚者用の再着色画像提示方法
映像情報メディア学会学生優秀発表賞, 野口展;佐藤俊治

Sep. 2021
神経回路学会
視覚数理モデルシミュレーションの高速化と錯視画像の探索
日本神経回路学会・優秀研究賞
Japan society

Sep. 2019
日本神経回路学会, https://www.uec.ac.jp/news/prize/2019/20190911_2050.html
A border-ownership model based on computational electromagnetism
日本神経回路学会・論文賞
Japan society

Sep. 2017
日本神経回路学会, 若手研究者（35歳以下）に対する表彰．中村大樹・佐藤俊治による研究に対して，中村氏に対して贈られた．
計算論的に最適な速度推定器
日本神経回路学会・大会奨励賞
Japan society

Oct. 2016
日本基礎心理学会
回転中心動揺錯視
第8回錯視・錯聴コンテスト入賞
Japan society

Dec. 2015
ロボットビジネス推進協議会
RTミドルウェア普及貢献賞
Japan society

Sep. 2015
日本神経回路学会
ＭＴ細胞の電気生理実験結果に関する計算論的再考察
日本神経回路学会・優秀研究賞

Dec. 2014
ロボットビジネス推進協議会
視覚脳科学研究を目的としたRTミドルウェアの応用と結果
RTミドルウェアコンテスト2014 奨励賞 日本ロボット工業会賞, 中村大樹;佐藤俊治;韓雪花;占部一輝
Japan society

Dec. 2014
ロボットビジネス推進協議会
視覚脳科学研究を目的としたRTミドルウェアの応用と結果
RTミドルウェアコンテスト2014 奨励賞 ベストサポート賞, 中村大樹;佐藤俊治;韓雪花;占部一輝
Japan society

Aug. 2014
MIT (Massachusetts Institute of Technology) Saliency Benchmarking team
The 1st prize for the total score of MIT Saliency Benchmarking
International academic award

Nov. 2010
APNNA Young Researcher Award, 佐藤俊治

Oct. 2009
理化学研究所理事長感謝状(野依良治理事長), 佐藤俊治

Sep. 2009
日本神経回路学会・論文賞, 佐藤俊治

Sep. 2006
日本神経回路学会・奨励賞, 佐藤俊治

両眼視野闘争の性質を利用した2色型色覚者用の再着色画像提示方法
野口 展; 佐藤 俊治
Last, 映像情報メディア学会論文誌, accepted (18-Apr-2024), 2024, Peer-reviwed
Scientific journal, Japanese

Depth Perception for the Image Displayed in Augmented-Reality (AR) Device
Nakajima Yutaka; Takemoto Masanori; Satoh Shunji
Transactions of the Virtual Reality Society of Japan, THE VIRTUAL REALITY SOCIETY OF JAPAN, 27, 2, 141-151, 30 Jun. 2022, The virtual image presented with Augmented Reality (AR) device is superimposed on the real world, and we can perceive it as if actually existing there. However, it is still unclear whether we could perceive a virtual image with the setting values for depth (distance and size). In this study, we examined the basic characteristics of depth perception for the AR images by the adjustment method (compared between an AR object and a real object) and the constant method (compared between an AR object and reference of distance). The results of experiments showed the overestimation of depth for a virtual image. In addition, the monocular observation would induce much overestimation than the binocular observation, while the sensitivity of depth improved with occlusion for another side of the eye where the virtual image was not presented. These results would help more appropriate presentation of the virtual image in the AR devices.
Japanese

Illusory Oscillation of the Central Rotation Axis
Yutaka Nakajima; Shohei Kakuda; Shunji Satoh
i-Perception, SAGE PUBLICATIONS LTD, 10, 4, 1-17, 25 Jul. 2019, Peer-reviwed, In this study, we report a novel visual illusion for rotational motion, in which the central rotation axis of a partially invisible (apparent) square is perceived as exhibiting oscillatory rotation. To investigate the cause of this illusion, we measured the central position of a static apparent shape using an adjustment method (Experiment 1) and manipulated the speed of the rotating apparent square to test whether the illusion could be cancelled out by counteracting rotation using a constant method (Experiment 2). The results revealed that the perceived central position of a static apparent shape was shifted toward the outside. The shifted position depended on the orientation of the stimulus, and its position was arranged as if it was moving in a circular trajectory. In addition, the cancellation technique using counteracting rotation was successful, and cancellation of faster rotation required a greater radius of counteracting rotation. These results indicated that the illusion is induced by an interaction between illusory shifts of the central position of the static shape and the summation of motion vectors or motion momentum (e.g., centrifugal force) derived from shape representation by perceptual completion.
Scientific journal, English

Computational study of depth completion consistent with human bi-stable perception for ambiguous figures
Eiichi Mitsukura; Shunji Satoh
Neural Networks, Elsevier Ltd, 99, 42-55, 01 Mar. 2018, Peer-reviwed, We propose a computational model that is consistent with human perception of depth in “ambiguous regions,” in which no binocular disparity exists. Results obtained from our model reveal a new characteristic of depth perception. Random dot stereograms (RDS) are often used as examples because RDS provides sufficient disparity for depth calculation. A simple question confronts us: “How can we estimate the depth of a no-texture image region, such as one on white paper?” In such ambiguous regions, mathematical solutions related to binocular disparities are not unique or indefinite. We examine a mathematical description of depth completion that is consistent with human perception of depth for ambiguous regions. Using computer simulation, we demonstrate that resultant depth-maps qualitatively reproduce human depth perception of two kinds. The resultant depth maps produced using our model depend on the initial depth in the ambiguous region. Considering this dependence from psychological viewpoints, we conjecture that humans perceive completed surfaces that are affected by prior-stimuli corresponding to the initial condition of depth. We conducted psychological experiments to verify the model prediction. An ambiguous stimulus was presented after a prior stimulus removed ambiguity. The inter-stimulus interval (ISI) was inserted between the prior stimulus and post-stimulus. Results show that correlation of perception between the prior stimulus and post-stimulus depends on the ISI duration. Correlation is positive, negative, and nearly zero in the respective cases of short (0–200 ms), medium (200–400 ms), and long ISI (>
400 ms). Furthermore, based on our model, we propose a computational model that can explain the dependence.
Scientific journal, English

A border-ownership model based on computational electromagnetism
Zaem Arif Zainal; Shunji Satoh
Neural Networks, Elsevier Ltd, 99, 114-122, 01 Mar. 2018, Peer-reviwed, The mathematical relation between a vector electric field and its corresponding scalar potential field is useful to formulate computational problems of lower/middle-order visual processing, specifically related to the assignment of borders to the side of the object: so-called border ownership (BO). BO coding is a key process for extracting the objects from the background, allowing one to organize a cluttered scene. We propose that the problem is solvable simultaneously by application of a theorem of electromagnetism, i.e., “conservative vector fields have zero rotation, or “curl.” We hypothesize that (i) the BO signal is definable as a vector electric field with arrowheads pointing to the inner side of perceived objects, and (ii) its corresponding scalar field carries information related to perceived order in depth of occluding/occluded objects. A simple model was developed based on this computational theory. Model results qualitatively agree with object-side selectivity of BO-coding neurons, and with perceptions of object order. The model update rule can be reproduced as a plausible neural network that presents new interpretations of existing physiological results. Results of this study also suggest that T-junction detectors are unnecessary to calculate depth order.
Scientific journal, English

Simple speed estimators reproduce MT responses and identify strength of visual illusion
Daiki Nakamura; Shunji Satoh
Neural Computing and Applications, Springer London, accepted, 7, 1-13, 02 Nov. 2017, Peer-reviwed, Computational models of vision should not only be able to reproduce experimentally obtained results
such models should also be able to predict the input–output properties of vision. Conventional models of MT neurons are based on the concept of velocity filtering, as proposed by Simoncelli and Heeger (Vis Res 38(5):743–761, 1998). As this report describes, we provide another interpretation of the computational function of MT neurons. An MT neuron can be a simple speed estimator with an upper limitation for correct estimation. Subsequently, we assess whether the MT model can account for illusory perception of “rotating drift patterns,” by which humans perceive illusory rotation (clockwise or counterclockwise rotation) depending on the background luminance. Moreover, to predict whether a pattern causes visual illusion, or not, we generate an enormous set of possible visual patterns as inputs to the MT model: (Formula presented.). Numerical quantities of model outputs obtained through a computer simulation for 88 inputs were used to estimate human illusory perception. Results of psychophysical experiments demonstrate that the model prediction is consistent with human perception.
Scientific journal, English

Formulation of Border-Ownership Assignment in Area V2 as an Optimization Problem
Zaem Arif Zainal; Shunji Satoh
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer Verlag, 10636, 859-866, 2017, Peer-reviwed, Border-ownership (BO) assignment, or the assignment of borders to an occluding object, is a primary step in visual perception. Physiological experiments have revealed the existence of neurons in area V2 that respond selectively to objects placed on a specific side of their response field. Although existing models can reproduce this phenomenon, they are not based on a clear computational theory. For this study, we formulated BO assignment as a well-defined optimization problem. We hypothesize that information related to BO assignment can be expressed as a conservative vector field. This conservative vector field is proposed as the gradient of a scalar field that carries information related to the depth order of the overlapping object. Conservative vector fields have zero curl (rotation). Using this theorem, we construct and solve an optimization problem. Numerical simulations demonstrate that a model based on our derived algorithm solves BO assignment for problems of perceived order and occlusion. Deduced neural networks provide insight into possible characteristics of lateral connections in area V2.
International conference proceedings, English

Application of RT-middleware to Computational Modelling of the Brain System
占部一輝; 中村大樹; 佐藤俊治; 韓雪花
計測自動制御学会論文誌, The Society of Instrument and Control Engineers, 52, 5, 264-275, 2016, Peer-reviwed, The brain is a typical complex system that executes visual information analysis, motor control, selective allocation of memories, and so on. We have developed a software platform to simulate the complex brain system numerically by computational models, especially focused on vision. Our platform for vision simulation bases on RT-middleware and OpenRTM-aist, which is a software platform to develop robotic system. A new datatype as a common interface of various vision models is provided. The new datatype and our software library enable automatic switching of transformation method of input/output data between vision models, i.e., shared memory or via computer network. We also provide a software package named by OpenCV-RTC which converts a lot of image processing functions of OpenCV into RT-components executable on OpenRTM-aist. We show that a novel model for estimation of fixation location of humans' eye is efficiently developed on our platform by parallel connection of existing models for eye fixation, and show that the new model significantly outperforms the existing models.
Scientific journal, Japanese

A Simple Visual Model Accounts for Drift Illusion and Reveals Illusory Patterns
Daiki Nakamura; Shunji Satoh
NEURAL INFORMATION PROCESSING, ICONIP 2016, PT IV, SPRINGER INT PUBLISHING AG, 9950, 9950, 191-198, 2016, Peer-reviwed, Computational models of vision should not only be able to reproduce experimentally obtained results; such models should also be able to predict the input-output properties of vision. We assess whether a simple computational model of neurons in the Middle Temporal (MT) visual area proposed by the authors can account for illusory perception of "rotating drift patterns," by which humans perceive illusory rotation (clockwise or counterclockwise) depending on the background luminance. Moreover, to predict whether a pattern causes visual illusion or not, we generate an enormous set of possible visual patterns as inputs to the MT model: 8(8) = 16,777,216, possible input patterns. Numerical quantities of model outputs by computer simulation for 8(8) inputs were used to estimate human illusory perception. Using psychophysical experiments, we show that the model prediction is consistent with human perception.
International conference proceedings, English

Unifying Computational Models for Visual Attention Yields Better Scores than State-of-the-art Models
Xuehua HAN; Shunji SATOH; Daiki NAKAMURA; Kazuki URABE
Advances in Neuroinformatics, 10.14931/aini2014.rii.6, RII-6, 27 Nov. 2014, Peer-reviwed
International conference proceedings, English

Improvement of Simulation Platform for providing reliable and easy use model simulation environment
Hidetoshi Ikeno; Tadashi Yamazaki; Yoshihiro Okumura; Shunji Satoh; Yoshimi Kamiyama; Yutaka Hirata; Keiichiro Inagaki; Akito Ishihara; Takayuki Kannon; Hiroaki Wagatsuma; Shiro Usui
Frontiers of Neuroinformatics, 10.3389/conf.fninf.2014.08.000, Feb. 2014, Peer-reviwed
International conference proceedings, English

A novel computational theory of MT neurons : Do MT neurons actually prefer t heir 'preferred speeds'?
Daiki Nakamura; Shunji Satoh
Neuro2013, accepted, Jun. 2013, Peer-reviwed
International conference proceedings, English

Computational study of depth perception for an ambiguous image region: How can we estimate the depth of black or white paper?
Eiichi Mitsukura; Shunji Satoh
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer, 8228, 3, 225-232, 2013, Peer-reviwed, We propose a new computational model that accounts for human perception of depth for "ambiguous regions," in which no information exists to estimate binocular disparity as seen in black and white papers. Random dot stereograms are widely used examples because these patterns provide sufficient information for disparity calculation. Then, a simple question confronts us: "how can we estimate the depth of non-textured images, like those on white paper?" In such non-textured regions, mathematical solutions of the spatial disparities are not unique but indefinite. We examine a mathematical description of depth estimation that is consistent with psychological experiments for non-textured images. Using computer simulation, we show that resultant depth-maps using our model based on the mathematical description above qualitatively reproduce human depth perception. © Springer-Verlag 2013.
International conference proceedings, English

Signal processing tolerant to noise accounts for human speed visual perception: (a counter-proposal against Bayesian models)
Shunji Satoh
The 22nd Annual Conference of Japanese Neural Network Society, xxx, Sep. 2012, Peer-reviwed
International conference proceedings, English

Computational identity between digital image inpainting and filling-in process at the blind spot
Shunji Satoh
NEURAL COMPUTING & APPLICATIONS, SPRINGER, 21, 4, 613-621, Jun. 2012, Peer-reviwed, Digital image inpainting (DII) is a computer algorithm that restores missing information of images such as those of old oil paintings. This problem occurs in human visual systems as well: we have blind spots (BS), but we see natural patterns in the BS region. This article presents the computational identity between the DII algorithm and the vision model for the filling-in process at the BS. Based on physiological evidence and conjecture, we define an evaluation function that evaluates the quality of filled-in (or inpainted) images. The definition of the evaluation function helps the original DII algorithm to improve the convergence speed. Numerical experiments demonstrate that the convergence speed using the energy function is three times faster than the original DII algorithm. Results show that the resultant filled-in patterns by the visual model are comparable with those of the DII algorithm.
Scientific journal, English

Collaborative software platform for computational brain research based on OpenRTM
Kazuki Urabe; Shunji Satoh; Taihei Kitagawa
6TH INTERNATIONAL CONFERENCE ON SOFT COMPUTING AND INTELLIGENT SYSTEMS, AND THE 13TH INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT SYSTEMS, IEEE, 326-331, 2012, Peer-reviwed, Many computational models should be integrated and executed on computers to simulate human visual systems. However, such integration has been difficult because every researcher has used different programming languages and different I/O formats. To solve this problem, we propose a software platform on which many vision models can be combined with less effort. The platform is based on "OpenRTM". Using our platform enables connection and execution of computational models written in C/C++, Java, Python, and MATLAB in a distributed/parallel computer environment.
International conference proceedings, English

Parallel Numerical Simulation of Visual Neurons for Analysis of Optical Illusion
Akira Egashira; Shunji Satoh; Hidetsugu Irie; Tsutomu Yoshinaga
2012 THIRD INTERNATIONAL CONFERENCE ON NETWORKING AND COMPUTING (ICNC 2012), IEEE, 130-136, 2012, Peer-reviwed, Detailed mechanism of optical illusion caused by visual neurons in human brain has not been well understood, and its numerical simulation is helpful to analyze visual system of humans. This paper describes implementation techniques of parallel numerical simulation to help understanding optical illusion by using a GPU-accelerated PC cluster. Our parallel acceleration techniques include following three points. Firstly, input images of the numerical simulation is efficiently calculated by dividing it images for multiple computation nodes using MPI (Message Passing Interface). Secondly, convolution, which is dominated computation for the optical flow, is accelerated by GPU. Finally, an algorithm to compute convolution specified to analyze optical illusion is proposed to speed up the simulation. Our experimental results show an interesting insight that values of optical flow for images causing optical illusion are quite different compared to that does not cause the optical illusion. We also demonstrate that our implementation of simulation works efficiently on the GPU-accelerated PC cluster.
International conference proceedings, English

Improvement of Simulation Platform for providing reliable and easy use model simulation environment
Hidetoshi Ikeno; Tadashi Yamazaki; Yoshihiro Okumura; Shunji Satoh; Yoshimi Kamiyama; Yutaka Hirata; Keiichiro Inagaki; Akito Ishihara; Takayuki Kannon; Hiroaki Wagatsuma; Shiro Usui
INCF Neuroinformatics, accepted, 2012, Peer-reviwed
International conference proceedings, English

Simulation Platform: Quick and easy access environment of model simulation in computational neuroscience
Hidetoshi Ikeno; Tadashi Yamaaki; Yoshihiro Okumura; Shunji Satoh; Yoshimi Kamiyama; Yutaka Hirata; Keiichiro Inagaki; Akito Ishihara; Takayuki Kannon; Shiro Usui
4th INCF Congress of Neuroinformatics, accepted, Dec. 2011, Peer-reviwed
International conference proceedings, English

Computational Model Resolving a Paradox between Speed Perception and Neural Property of MT Area
Shunji Satoh
JNNS 2011, accepted, Dec. 2011, Peer-reviwed
International conference proceedings, English

A large-scale whole visual system model integrated by PLATO and its implementation on high performance computer
Keiichiro Inagaki; Takayuki Kannon; Yoshimi Kamiyama; Shunji Satoh; Nilton Kamiji; Daiki Sone; Kazuki Urabe; Shiro Usui
4th INCF Congress of Neuroinformatics, accepted, Dec. 2011, Peer-reviwed
International conference proceedings, English

Reprint of: Simulation Platform: a cloud-based online simulation environment.
Yamazaki T; Ikeno H; Okumura Y; Satoh S; Kamiyama Y; Hirata Y; Inagaki K; Ishihara A; Kannon T; Usui S
Neural networks : the official journal of the International Neural Network Society, 24, 9, 927-932, Nov. 2011, Peer-reviwed
Scientific journal

Simulation Platform: A cloud-based online simulation environment
Tadashi Yamazaki; Hidetoshi Ikeno; Yoshihiro Okumura; Shunji Satoh; Yoshimi Kamiyama; Yutaka Hirata; Keiichiro Inagaki; Akito Ishihara; Takayuki Kannon; Shiro Usui
NEURAL NETWORKS, PERGAMON-ELSEVIER SCIENCE LTD, 24, 7, 693-698, Sep. 2011, Peer-reviwed, For multi-scale and multi-modal neural modeling, it is needed to handle multiple neural models described at different levels seamlessly. Database technology will become more important for these studies, specifically for downloading and handling the neural models seamlessly and effortlessly. To date, conventional neuroinformatics databases have solely been designed to archive model files, but the databases should provide a chance for users to validate the models before downloading them. In this paper, we report our on-going project to develop a cloud-based web service for online simulation called "Simulation Platform". Simulation Platform is a cloud of virtual machines running GNU/Linux. On a virtual machine, various software including developer tools such as compilers and libraries, popular neural simulators such as GENESIS, NEURON and NEST, and scientific software such as Gnuplot, R and Octave, are pre-installed. When a user posts a request, a virtual machine is assigned to the user, and the simulation starts on that machine. The user remotely accesses to the machine through a web browser and carries out the simulation, without the need to install any software but a web browser on the user's own computer. Therefore, Simulation Platform is expected to eliminate impediments to handle multiple neural models that require multiple software. (C) 2011 Elsevier Ltd. All rights reserved.
Scientific journal, English

Simulation Platform: Cloud-computing meets computational neuroscience
T Yamazaki; H Ikeno; Y Okumura; S Satoh; Y Hirata; A Ishihara; K Kannon; S Usui
Proc. of CNS 2011, in print, Jul. 2011, Peer-reviwed
International conference proceedings, English

Multi-GPU acceleration of optical flow computation in visual functional simulation
Junichi Ohmura; Akira Egashira; Shunji Satoh; Takefumi Miyoshi; Hidetsugu Irie; Tsutomu Yoshinaga
Proceedings - 2011 2nd International Conference on Networking and Computing, ICNC 2011, IEEE Computer Society, accepted, 228-234, 2011, Peer-reviwed, Numerical simulation for visual processing of the human brain is one of time-consuming applications. This paper shows acceleration techniques for a simulation program of the visual processing. We parallelize convolution calculations, which are core operations, which the simulation program requests, on a GPU-accelerated PC cluster. Our implementation includes three improvement points. Firstly, we consider efficient data mapping onto global and shared memories1 of the GPU. Secondly, multiple convolutions for the same input data are computed by each node's GPU, referred to as package execution. Finally, an input 2-dimensional image is divided into regions and convolutions for these regions are executed in parallel utilizing MPI (Message Passing Interface). Our experimental results show a linear speedup up to 12 nodes in the PC cluster for the convolution program. We also show the effects of the package execution and reduced communication on NVIDIA tesla C1060 and C2070, respectively. © 2011 IEEE.
International conference proceedings, English

Environment for an integrative model simulation: PLATO
Keiichiro Inagaki; Takayuki Kannon; Yoshimi Kamiyama; Shunji Satoh; Nilton Kamiji; Yutaka Hirata; Akito Ishihara; Hayaru Shouno; Shiro Usui
Proc. of Neuroinformatics 2010, 1, ???, Aug. 2010, Peer-reviwed
International conference proceedings, English

Simulation Platform: Model simulation on the cloud
Shiro Usui; Tadashi Yamazaki; Hidetoshi Ikeno; Okumura Yoshihiro; Shunji Satoh; Kamiyama Yoshimi; Hirata Yutaka; Inagaki Keiichiro; Ishihara Akito; Kannon Takayuki; Kamiji Nilton; Akazawa Fumihiko
Proc. of Neuroinformatics 2010, 1, ???, Aug. 2010, Peer-reviwed
International conference proceedings, English

Efficient Representation by Horizontal Connection in Primary Visual Cortex
Hiroaki Sasaki; Shunji Satoh; Shiro Usui
NEURAL INFORMATION PROCESSING: THEORY AND ALGORITHMS, PT I, SPRINGER-VERLAG BERLIN, 6443, 132-+, 2010, Peer-reviwed, Neurons in the primary visual cortex (V1) encode natural images that are exposed. As a candidate encoding principle, the efficient coding hypothesis was proposed by Attneave (1954) and Barlow (1961). This hypothesis emphasizes that the primary role of neurons in the sensory area is to reduce the redundancy of the external signal and to produce a statistically efficient representation. However, the outputs of neurons in V1 are statistically dependent because their classical receptive fields largely overlap and natural images have structures such as edges and textures. As described in this paper, we propose that the computational role of horizontal connections (HCs) is to decrease statistical dependency and attempt to self-organize the spatial distribution of HCs from natural images. In addition, we show that our neural network model with self-organized HCs can reproduce some nonlinear properties of V1 neurons, e.g. size-tuning and contextual modulation. These results support the efficient coding hypothesis and imply that HCs serve an important role in decreasing statistical dependency in V1.
International conference proceedings, English

Kalman filter model can explain the temporal receptive field of motion selective V1 neurons
Shunji Satoh; Yutaka Sakaguchi; Hiroaki Sasaki; Shiro Usui
NEUROSCIENCE RESEARCH, ELSEVIER IRELAND LTD, 68, E379-E380, 2010, Peer-reviwed
International conference proceedings, English

Kalman filter model can explain the temporal receptive field of motion selective V1 neurons
Shunji Satoh; Yutaka Sakaguchi; Hiroaki Sasaki; Shiro Usui
NEUROSCIENCE RESEARCH, ELSEVIER IRELAND LTD, 68, E379-E380, 2010, Peer-reviwed
International conference proceedings, English

Neural implementation of coarse-to-fine processing in V1 simple neurons
Hiroaki Sasaki; Shunji Satoh; Shiro Usui
NEUROCOMPUTING, ELSEVIER SCIENCE BV, 73, 4-6, 867-873, Jan. 2010, Peer-reviwed, Coarse-to-fine processing has been observed in various areas of the visual cortex. For example, some receptive fields (RFs) of neurons in the primary visual cortex (V1) shrink spatially as time progresses. Such VI neurons become more sensitive to higher spatial period stimulus in 20 ms. Furthermore, orientation selectivity in VI also increases, that is, orientational coarse-to-fine processes. As described herein, we investigate the neural substances related to coarse-to-fine processing. We find that such coarse-to-fine processing corresponds to deblurring operations to achieve V1 neural output with spatially and orientationally higher resolutions. We show computationally that the short-range horizontal connections (SHCs) realize the deblurring operation.
We Simulate a V1 network model with SHCs based on the Sasaki model [1] (Sasaki and Satoh, 2009). The shrinking VI receptive-field and increased orientational selectivity are caused by neural deblurring operations through SHCs. The model properties are qualitatively consistent with physiological data. (C) 2009 Published by Elsevier B.V.
Scientific journal, English

Super resolution: Another computational role of short-range horizontal connection in the primary visual cortex
Hiroaki Sasaki; Shunji Satoh
NEURAL NETWORKS, PERGAMON-ELSEVIER SCIENCE LTD, 22, 4, 362-372, May 2009, Peer-reviwed, Recent physiological data related to the primary visual cortex (VI) have shown various contextual effects in the non-classical receptive field (nCRF). Contextual modulation, size tuning and altered sensitivity of orientation are typical examples of such contextual effects in the nCRF. These phenomena in the nCRF have been thought to be caused by short-range horizontal connection (SHC). However, SHC does not necessarily contribute only to these phenomena. These phenomena might be merely secondary phenomena by the fundamental role of SHC. In this paper, we specifically address the overcomplete properties in VI. Then the fundamental role of SHC is examined from image-processing points of view. Super resolution is proposed as a strong candidate for the fundamental role of SHC. Super resolution is an engineering method that obtains a high-resolution image from a low-resolution image(s). The distribution of SHC is deductively derived by adopting a reverse diffusion technique, which is one of various available super-resolution techniques. The spatial distribution of our proposed SHC is isotropic on the orientation map. This characteristic is consistent with physiological data. In addition to that, contextual modulation, size tuning and altered sensitivity of orientation in numerical experiments using our proposed SHC can be reproduced qualitatively. These results indicate that these phenomena are secondary phenomena by super-resolution processing. (C) 2008 Elsevier Ltd. All rights reserved.
Scientific journal, English

Engineering-approach accelerates computational understanding of V1-V2 neural properties
Shunji Satoh; Shiro Usui
COGNITIVE NEURODYNAMICS, SPRINGER, 3, 1, 1-8, Mar. 2009, Peer-reviwed, We present two computational models (i) long-range horizontal connections and the nonlinear effect in V1 and (ii) the filling-in process at the blind spot. Both models are obtained deductively from standard regularization theory to show that physiological evidence of V1 and V2 neural properties is essential for efficient image processing. We stress that the engineering approach should be imported to understand visual systems computationally, even though this approach usually ignores physiological evidence and the target is neither neurons nor the brain.
Scientific journal, English

A Next Generation Modeling Environment PLATO: Platform for Collaborative Brain System Modeling
Shiro Usui; Keiichiro Inagaki; Takayuki Kannon; Yoshimi Kamiyama; Shunji Satoh; Nilton L. Kamiji; Yutaka Hirata; Akito Ishihara; Hayaru Shouno
NEURAL INFORMATION PROCESSING, PT 1, PROCEEDINGS, SPRINGER-VERLAG BERLIN, 5863, 84-+, 2009, Peer-reviwed, To understand the details of brain function, a large scale system model that reflects anatomical and neurophysiological characteristics needs to be implemented. Though numerous computational models of different brain areas have been proposed, these integration for the development of a large scale model have not yet been accomplished because these models were described by different programming languages, and mostly because they used different data formats. This paper introduces a platform for a collaborative brain system modeling (PLATO) where one can construct computational models using several programming languages and connect them at the I/O level with a common data format. As an example, a whole visual system model including eye movement, eye optics, retinal network and visual cortex is being developed. Preliminary results demonstrate that the integrated model successfully simulates the signal processing flow at the different stages of visual system.
International conference proceedings, English

Computational theory and applications of a filling-in process at the blind spot
Shunji Satoh; Shiro Usui
NEURAL NETWORKS, PERGAMON-ELSEVIER SCIENCE LTD, 21, 9, 1261-1271, Nov. 2008, Peer-reviwed, A mathematical model for filling-in at the blind spot is proposed. The general scheme of the standard regularization theory was used to derive the model deductively. First, we present the problems encountered with a diffusion equation, which is frequently used for various types of perceptual completion. To solve these problems, we investigated the computational meaning of a neural property discovered by Matsumoto and Komatsu [Matsumoto, M., &, Komatsu, H. (2005). Neural responses in the macaque V1 to bar stimuli with various lengths presented on the blind spot. Journal of Neurophysiology, 93, 2374-2387). Based on our observations, we introduce two types of curvature information of image properties into the a priori knowledge of missing images in the blind spot. Moreover, two different information pathways for filling-in, which were Suggested by results of physiological experiments (slow conductive paths of horizontal connections in V1, and fast feedforward/feedback paths via V2), were considered theoretically as the neural embodiment of in adiabatic approximation between V1 and V2 interaction. Numerical simulations show that the output of the proposed model for filling-in is consistent with neurophysiological experimental results. The model can be used as a powerful tool for digital image inpainting. (C) 2008 Elsevier Ltd. All rights reserved.
Scientific journal, English

Engineering-approach accelerates computational understanding of V1-V2 neural properties
Shunji Satoh; Shiro Usui
NEURAL INFORMATION PROCESSING, PART I, SPRINGER-VERLAG BERLIN, 4984, 1051-1060, 2008, Peer-reviwed, We present two computational models: (i) long-range horizontal connections and the nonlinear effect in V1 and (ii) the filling-in process at the blind spot. Both models are obtained deductively from standard regularization theory to show that physioligical evidence of V1 and V2 neural properties is essential for efficient image processing. We stress that the engineering approach should be imported to understand visual systems computationally, even though this approach usually ignores physiological evidence and the target is neither neurons nor the brain.
International conference proceedings, English

Computational understanding and modeling of filling-in process at the blind spot
Shunji Satoh; Shiro Usui
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer, 4984, 1, 943-952, 2008, A visual model for filling-in at the blind spot is proposed. The general scheme of standard regularization theory is used to derive a visual model deductively. First, we indicate problems of the diffusion equation, which is frequently used for various kinds of perceptual completion. Then, we investigate the computational meaning of a neural property discovered by Matsumoto and Komatsu (J. Neurophysiology, vol. 93, pp. 2374-2387, 2005) and introduce second derivative quantities related to image geometry into a priori knowledge of missing images on the blind spot. Moreover, two different information pathways for filling-in (slow conductive paths of horizontal connections in V1, and fast feedforward/feedback paths via V2) are regarded as the neural embodiment of adiabatic approximation between V1 and V2 interaction. Numerical simulations show that the outputs of the proposed model for filling-in are consistent with a neurophysiological experimental result, and that the model is a powerful tool for digital image inpainting. © 2008 Springer-Verlag Berlin Heidelberg.
International conference proceedings, English

Physiologically plausible and theoretically optimal model of V1 receptive fields
Shunji Satoh; Shiro Usui
Proc of NBNI-2008, *, *, 2008
International conference proceedings, English

Psychophysical analyses of the size effects of spatial attention on figure-ground assignment
Souta Hidaka; Shunji Satoh; Jiro Gyoba
Fechner Day, *, *, 2007, Peer-reviwed
International conference proceedings, English

A visual model for object detection based on active contours and level-set method
Shunji Satoh
BIOLOGICAL CYBERNETICS, SPRINGER, 95, 3, 259-270, Sep. 2006, Peer-reviwed, A visual model for object detection is proposed. In order to make the detection ability comparable with existing technical methods for object detection, an evolution equation of neurons in the model is derived from the computational principle of active contours. The hierarchical structure of the model emerges naturally from the evolution equation. One drawback involved with initial values of active contours is alleviated by introducing and formulating convexity, which is a visual property. Numerical experiments show that the proposed model detects objects with complex topologies and that it is tolerant of noise. A visual attention model is introduced into the proposed model. Other simulations show that the visual properties of the model are consistent with the results of psychological experiments that disclose the relation between figure-ground reversal and visual attention. We also demonstrate that the model tends to perceive smaller regions as figures, which is a characteristic observed in human visual perception.
Scientific journal, English

A Visual Model for Object Detection Based on the Computational Principle of Active Contour
SATOH Shunji; MIYAKE Shogo
The transactions of the Institute of Electronics, Information and Communication Engineers. D-II, 社団法人電子情報通信学会, 88, 7, 1257-1268, Jul. 2005, Peer-reviwed, 背景から物体(図)領域を検出する方法を提案し, 提案方法を実現する神経回路網モデルを提案する.提案方法は, 画像工学の立場から提唱された動的輪郭法の計算原理に基づいており, 物体の境界がもつ幾何的特徴, 及び認知心理学的知見を統合することで得られる.本研究ではまず, 動的輪郭法の定常状態に関する解析をした後, 物体領域検出のためのエネルギーを定義し, エネルギー最小化の原理から神経回路網モデルの結合や動作を導出する.本研究では神経回路網モデルの提案のみならず, 動的輪郭法の重要な問題点を解決し, 新しい物体検出法の提案も行う.数値実験により, (i)トポロジーが複雑な物体, (ii)複数の物体の検出に成功し, (iii)ノイズに頑健であることを示す.提案モデルは, 神経生理学的知見と一致し, 図地反転現象を説明するモデルであることも示す.
Japanese

New Neocognitron-Type Network and Its Learning Method Based on ICA and PCA
SHIMOMURA Masao; SATOH Shunji; MIYAKE Syogo; ASO Hirotomo
The transactions of the Institute of Electronics, Information and Communication Engineers. D-II, 社団法人電子情報通信学会, 88, 4, 769-777, Apr. 2005, Peer-reviwed, ネオコグニトロンは, 高い認識率と拡張性をもつパターン認識用の階層型ニューラルネットワークであるが, その性能を引き出すためには多数存在するパラメータを認識対象に応じて適切に調整する必要があった.そこで本論文では, ネオコグニトロンの各階層で行われている処理が次元圧縮である点に着目し, 統計的な次元圧縮法である主成分分析(PCA), 独立成分分析(ICA)及び部分空間法をネットワークの学習法として導入することで, パラメータ数の削減とパラメータ変動への頑健性向上を図る.また, これらの手法を導入するためにネオコグニトロンを簡略化したネットワークを提案する.手書き数字及び顔画像データベースを用いた認識実験により, 提案する学習法がパラメータの変動に頑健で, かつ認識対象によらず同じパラメータで適切な学習が行えることを検証した.
Scientific journal, Japanese

A model of overt visual attention based on scale-space theory
Shunji Satoh; Shogo Miyake
Systems and Computers in Japan, 35, 10, 1-13, Sep. 2004, Peer-reviwed, This paper proposes a vision model for object detection based on scale-space theory, considering knowledge obtained in neurophysiology and human visual characteristics obtained in visual psychology. The proposed model is principally composed of (a) the early vision model and (b) the attention calculation model. In this paper, it is shown first that the visual characteristics can be described by a discretized scale space, considering their multichannel property, spatial nonuniformity, and orientation selectivity. The early vision model is formulated. Next, the attention calculation model and the operation algorithm are formulated on the basis of the knowledge obtained by scale-space theory. The numerical experiments reveal that the proposed model has the following properties, (i) For objects with a high intensity difference, the center of overt attention moves to the center of the object, (ii) The spatial extent of overt/covert attention can be calculated adequately, (iii) The object is captured at the central region of the retina, where the resolution is the highest. Since the proposed model is based on scale-space theory, the theory or model can be easily extended. There are other advantages from an engineering standpoint, such as simple structure, easy implementation, small computational requirements, and very few parameters to be adjusted. © 2004 Wiley Periodicals, Inc.
Scientific journal, English

A Model of Overt Visual Attention Based on Scale-Space Theory
SATOH Shunji; MIYAKE Shogo
The transactions of the Institute of Electronics, Information and Communication Engineers. D-II, 社団法人電子情報通信学会, 86, 10, 1490-1501, Oct. 2003, Peer-reviwed, 神経生理学で得られた知見や,視覚心理学で得られたヒトの視覚特性を考慮した,スケールスペース理論に基づく物体検出のための視覚モデルを提案する.提案モデルは主に(a)初期視覚モデル,(b)注意計算モデルで構成される.本論文ではまず,多重チャネル性,空間不均一性,及び方位選択性を考慮した視覚特性が,離散化されたスケールスペースで記述できることを示し,初期視覚モデルの定式化を行う.次に,スケールスペース理論で得られた知見に基づき,注意計算モデルの定式化,及び動作アルゴリズムの定式化を行う.数値実験により提案モデルは次の性質をもつことがわかった.(I)輝度差の大きい物体に対してはその物体の中心位置へ視点を移動し,(ii)適切な注意/注視範囲を計算し,(iii)最も解像度が高い中心部に物体をとらえる.提案モデルはスケールスペース理論に基づくため理論的な考察やモデルの拡張が容易である.更に,構造が単純で実装が容易,計算量が小さい,調節するパラメータ数が極少数である等の工学的利点もある.
Japanese

A New Learning Method of a Neocognitron-type Network Based on ICA and PCA
Masao Shimomura; Shunji Satoh; Shogo Miyake; Hirotomo Aso
Information Technology Letters, 2, 155-157, Sep. 2003, Peer-reviwed
Scientific journal, Japanese

A New Learning Method of a Hierarchical Network Using ICA and PCA for Image Recognition
Masao Shimomura; Shunji Satoh; Shogo Miyake; Hirotomo Aso
Proceeding of Artificial Neural Networks and Neural Information, ICANN/ICONIP 2003, 294-297, Aug. 2003, Peer-reviwed
International conference proceedings, English

A Model for Selective Visual Attention Based on Discrete Scale-Spaces.
Shunji Satoh; Shogo Miyake
Knowledge-Based Intelligent Information and Engineering Systems(KES), Springer, 147-154, 2003
International conference proceedings

A Neural Network Model for Pattern Recognition Based on Hypothesis and Verification with Moving Region of Attention
Masao Shimomura; Shunji Satoh; Shogo Miyake; Hirotomo Aso
Artificial Neurwl Networks - ICANN, Springer, 1275-1280, Aug. 2002, Peer-reviwed
International conference proceedings, English

Neocognitron-Type Network for Recognizing Rotated and Shifted Patterns with Reduction of Resources.
Shunji Satoh; Shogo Miyake; Hirotomo Aso
Connectionist Models of Neurons, Learning Processes and Artificial Intelligence, 6th International Work-Conference on Artificial and Natural Neural Networks, Springer, 215-222, 2001
International conference proceedings

階層的パターン統合処理に基づく視覚情報処理モデルに関する研究
佐藤俊治
東北大学審査博士学位論文, 2000

Rotation-invariant neocognitron
Shunji Satoh; Jousuke Kuroiwa; Hirotomo Aso; Shogo Miyake
Systems and Computers in Japan, 30, 4, 31-40, Apr. 1999, Peer-reviwed, A neocognitron is a hierarchical neural model which can recognize shifted patterns in positions and/or deformed patterns after unsupervised learning, but the model cannot recognize rotated patterns. We propose a rotation-invariant neocognitron which can also recognize rotated patterns by extending the function of the neocognitron. We also propose a new learning method, by which thresholds of the rotation-invariant neocognitron are controlled in a training phase so that the model shows the expected performance for the recognition of rotated patterns. We show that the model can recognize rotated patterns as well as shifted and/or deformed patterns in computer simulations. The model also acquires an ability of recognition of locally rotated patterns, which cannot be correctly recognized by the standard model, because the rotation-invariant neocognitron recognizes the entirety of an input pattern through multilayered processing of local features including rotational information.
Scientific journal, English

Pattern recognition system with top-down process of mental rotation
Shunji Satoh; Hirotoino Aso; Shogo Miyake; Jousuke Kuroiwa
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer Verlag, 1606, 816-825, 1999, Peer-reviwed, A new model which can recognize rotated, distorted, scaled, shifted and noised patterns is proposed. The model is constructed based on psychological experiments in a mental rotation. The model has two types of processes: (i) one is a bottom-up process in which pattern recognition is realized by means of a rotation-invariant neocognitron and a standard neocognitron and (ii) the other is a top-down process in which a mental rotation is executed by means of a model of associative recall in visual pattern recognition. In computer simulations, it is shown that the model can recognize rotated patterns without training those patterns.
International conference proceedings, English

RECOGNITION OF HAND-WRITTEN PATTERNS BY ROTATION-INVARIANT NEOCOGNITRON
Shunji Satoh; Jousuke Kuroiwa; Hirotomo Aso; Shogo Miyake
ICONIP'98 Proceedings, IOA Press, 1, 295-299, Oct. 1998, Peer-reviwed
Scientific journal, English

Rotation-Invariant Neocognitron
SATOH Shunji; KUROIWA Jousuke; ASO Hirotomo; MIYAKE Shogo
The transactions of the Institute of Electronics, Information and Communication Engineers, 社団法人電子情報通信学会, 81, 6, 1365-1374, Jun. 1998, Peer-reviwed, 教師なし学習でパターン認識が可能なネオコグニトロンは位置ずれや変形に頑強な認識をする階層型神経回路モデルであるが, 回転したパターンを認識することはモデルの構成上不可能であった.本論文では, 変形に対する頑強性を実現しているネオコグニトロンの構成をそのまま回転に対して拡張し, 回転パターンも認識可能な回転対応型ネオコグニトロンを提案する.更に, 回転対応型ネオコグニトロンが所期の機能を獲得するための学習法として, 学習段階でしきい値を変化させる.しきい値制御学習法を提案する.また, 実際に回転対応型ネオコグニトロンを計算機上で構築し, 平行移動や変形だけでなく回転パターンの認識が可能であることを示す.本モデルは回転の情報を含む部分特徴量の処理を多層化することでパターン全体を認識しているため, 従来手法では認識不可能であったパターンの一部が回転しているようなパターンも認識可能となった.
Japanese

Recognition of rotated patterns using neocognitron
Shunji Satoh; Jousuke Kuroiwa; Hirotomo Aso; Shogo Miyake
Progress Connectionist-Based Information Systems (Proc. of ICONIP'97), Springer, 1, 112-116, Sep. 1997, Peer-reviwed
Scientific journal, English

Recognition of Rotated Patterns Using Neocognitron
Shunji Satoh; Jousuke Kuroiwa; Hirotomo Aso; Shogo Miyake
Peer-reviwed, A rotation-invariant neocognitron is constructed by extending the neocognitron which can recognize translated, scaled and/or distorted patterns from training ones. In constructing the model two technical methods during the learning, a "threshold-controlling method" and a "rotation matrix method", are proposed. In numerical simulations, it is shown that the model can recognize globally and/or locally rotated patterns in an arbitrary angle without learning the patterns themselves. 1 Introduction It is very important to realize recognition systems insensitive to different kinds of scaling, translation, distortion and rotation. Many neural networks have been proposed to achieve this purpose. Neocognitron[Fukushima, 1988] is a multi-layered neural network model for pattern recognition which is considerably robust against distortion, scaling, and/or translation of patterns. After unsupervised learning, it can recognize input patterns without being affected by distortion, change in these siz...

Simple Color-Correction using a Mirror and a Virtual Color Chart
Yuki KOJIMA; Shunji Satoh
Last, Mar. 2024, 情報処理学会 第86回全国大会, Japanese, Summary national conference

Image recoloring/presentation method utilizing binocular rivalry for color-vision-deficiencies
Noguchi Hiraku; Satoh Shunji
Last, Sep. 2023, 映像情報メディア学会, Summary national conference

A computational model of the interaction between motion perception and color perception
Ishiyama Yasuo; Satoh Shunji
Last, Sep. 2023, 日本神経回路学会全国大会, Japanese, Summary national conference

Visual-cortical mapping reproduced by the self-organization algorithm
Daisuke Sano; Shunji Satoh
Last, Mar. 2023, IEICE technical report

Perception of curved surface depending on viewing distance
Aoshima Hatsuho; Satoh Shunji
We investigated the perceptual difference of cylindrically curved surfaces of random-dot-stereograms (RDS) and real objects. The ‘real’ objects used in this study are transparent sheet filled with with random-dots, and the objects are illuminated by the back light emitted from a LED monitor. In experiments of this study, human subjects observed cylinders of five kinds of radii with three different viewing distances. Experimental results indicate that the perceptual difference of depth between RDSs and real objects depends on the viewing distance and the visual angle of objects. Moreover, we find a mathematically simple description of perceived surface, viewing distance and visual angle of objects. As an application of virtual reality, we propose a method computing horizontal difference of RDSs which could be perceived as the same surfaces of real objects., The Institute of Image Information and Television Engineers, 2021, ITE Technical Report, 41.08, 17-20, Japanese, 1342-6893, 2424-1970, 130008043738

自己運動中の移動物体知覚特性の計測と計算論的考察—Experimental and computational study of visual perception for moving objects during self-motion—ヒューマン情報処理
成田 侑毅; 佐藤 俊治
電子情報通信学会, Oct. 2019, 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 119, 254, 41-45, Japanese, 0913-5685, 40022064231, AA1123312X

Depth perception for the image displayed by augmented reality (AR) systems
Nakajima Yutaka; Kikuchi Masahiro; Satoh Shunji
The Japanese Psychological Association, 25 Sep. 2018, The Proceedings of the Annual Convention of the Japanese Psychological Association, 82, 1AM-057-1AM-057, Japanese, 2433-7609, 130007680382

Online simulation environment for computational neuroscience and data analysis
Ikeno H; Yamazaki T; Kannon T; Okumura Y; Kamiyama Y; Ishihara A; Inagaki K; Hirata Y; Satoh S; Wagatsuma H; Asai Y; Yamaguchi Y; Usui S
Aug. 2017, Neuroinformatics 2017, D1, Summary international conference

周辺視野における視覚情報処理に聴覚刺激が及ぼす影響—Effect of auditory stimulus on information processing in peripheral vision—ヒューマン情報処理
渡部 貴行; 佐藤 好幸; 佐藤 俊治
電子情報通信学会, Mar. 2017, 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 116, 513, 17-20, Japanese, 0913-5685, 40021161658, AA1123312X

曲面知覚の観察距離依存性に関する研究—Perception of curved surface depending on viewing distance—立体映像技術 ヒューマンインフォメーション
青島 初帆; 佐藤 俊治
映像情報メディア学会, Mar. 2017, 映像情報メディア学会技術報告 = ITE technical report, 41, 8, 17-20, Japanese, 1342-6893, 40021162379, AN1059086X

Applying an electromagnetic theorem to solve the Border-Ownership assignment problem: Border Ownership計算問題の電磁気学定理による定式化とモデル化
ザエム アリフ ザイナル; サトウ シュンジ
Mar. 2016, 115, 514, 107-112, English, 0913-5685, 40020791472, AA1123312X

Development of an on-line simulation platform for neuroscience research
Ikeno H; Yamazaki T; Kamiyama Y; Ishihara A; Inagaki K; Hirata Y; Okumura Y; Kannon T; Satoh S; Wagatsuma H; Asai Y; Yamaguchi Y; Usui S
Aug. 2015, Front. Neurosci. Conference Abstract: Neuroinformatics 2015

A simple visual model for optical-flow estimation generates illusory patterns
Nakamura Daiki; Satoh Shunji
Computational models of vision should not only be able to reproduce the Input-Output relationship of the human vision or neural systems, but also should those models be able to predict the I/O of vision. First, we examine if an MT model proposed by authors can reproduce the dependency of background luminance of rotating drift-illusory patterns. Moreover, to evaluate the plausibility of the MT model, we generate an enormous set of possible visual patterns; 8^8=16,777,216 patterns, as inputs to the MT model. The numerical quantities of model outputs by computer simulation for 8^8 inputs can be used as estimation of humans' perception. By psychophysical experiments, we show that the model prediction is consistent with human perception., The Institute of Electronics, Information and Communication Engineers, 16 Mar. 2015, IEICE technical report. ME and bio cybernetics, 114, 514, 151-156, Japanese, 0913-5685, 110010021869, AN1001320X

A novel computational concept of the physiological properties of MT neurons : Do MT neurons actually prefer to their 'preferred speeds'?
Nakamura Daiki; Satoh Shunji
Middle Temporal (MT) neurons are believed to have a preferred speed: the response curves of MT neurons illustrate unimodal functions with respect to image speeds. However, we found that our model, which is based on the Lucas-Kanade (LK) method, was also reproducing unimodal functions by numerical simulation. Interestingly, although the LK method does not have "preferred speed", it shows a unimodal function of its estimated speed with respect to input speed due to the existence of upper limitation of correct estimation. Therefore, we put forward a new notion that MT neurons just estimate image speeds not have preferred speed. Moreover, the change of tuning width of MT neurons affected by image contrast is also reproduced by our LK-based model of MT neurons., The Institute of Electronics, Information and Communication Engineers, 17 Mar. 2014, IEICE technical report. Neurocomputing, 113, 500, 41-46, Japanese, 0913-5685, 110009862331, AN10091178

ユニーク＆エキサイティング サイエンス
梶谷誠; 佐藤俊治; 他
Apr. 2013, 近代科学社, 1, Japanese, Introduction other

Multi-GPU Acceleration of Numerical Simulation for the Linear Model of Visual Neurons
Ohmura Junichi; Shunji Satoh; Akira Egashira; Takefumi Miyoshi; Hidetsugu Irie; Tsutomu Yoshinaga
A popular approach to understand our human visual "functions" is performing the computational simlutaions of its linear models created with a focus on input-output relations of cells. However, due to a lot of simulation time for a huge amount of cells, it often happen that only simplified models of selected visual "functions" have been dealt with. So, in order to speed up the simulation, in this study we attempt to parallelize the program and perform it on a GPU-accelerated PC cluster. We show a performance comparison between NVIDIA C1060 and C2070, which are difference in their architectur..., Information Processing Society of Japan (IPSJ), 20 Jul. 2011, IPSJ SIG Notes, 2011, 8, 1-8, Japanese, 110008583364, AN10463942

Parallel Numerical Simulation for the Linear Model of Visual Neurons with MPI
Yusuke Saito; Shunji Satoh; Ohmura Junichi; Takefumi Miyoshi; Hidetsugu Irie; Tsutomu Yoshinaga
Numerical simulation for the linear model of visual neurons is the most important approach to understand our visual system from computational viewpoints. We attempt to parallelize the time-consuming simulation on a cluster computer system. We achieved 43% reduction in simulation time by MPI implementation of spatio-temporal convolution formulated in the linear model. Moreover, by analyzing the simulation results, unknown factors on visual illusion are unveiled., Information Processing Society of Japan (IPSJ), 08 Mar. 2011, IPSJ SIG Notes, 2011, 4, 1-8, Japanese, 2186-2583, 110008583334, AN10463942

V1野の非線形な応答を説明するエッジ検出モデル
佐々木博昭; 佐藤俊治; 臼井支朗
第一次視覚野(V1野)の単純型細胞は，特定の位置に特定の方位をもったエッジを検出する．しかし工学的な問題として，位置精度と方位精度の両者を，限界値を超えて高めることはできない（不確定性原理）．そこで本研究では，逆拡散における超解像法を拡張・適用することで問題の解決を試みた．その結果、位置と方位に関して高精度なエッジ検出が可能となった。加えて構築されたモデルは、神経生理学的な現象を良く再現することもわかった., The Institute of Systems, Control and Information Engineers, 19 May 2010, システム制御情報学会研究発表講演会講演論文集(CD−ROM), 54th, ROMBUNNO.W32-3-86, Japanese, 201002281223789279, 130006985675

A computational theory of horizontal connection in primary visual cortex
SASAKI Hiroaki; SATOH Shunji; USUI Shiro
It is believed that in primary visual cortex(V1), neurons efficiently encode natural images by the sparse outputs of these neurons. However, such outputs of V1 cells are not sufficiently sparse because natural images have a spatial structure like edge and texture. Also the classical receptive fields of V1 cells are spatially overlapped. In this paper, we propose that the computational role of horizontal connection(HC) in V1 is to increase the efficiency of image coding. The estimated HC indicates a mexican hat like spatial distribution to natural images. First, we define a mathematical func..., The Institute of Electronics, Information and Communication Engineers, 02 Mar. 2010, IEICE technical report. Neurocomputing, 109, 461, 7-12, Japanese, 0913-5685, 110008004267, AN10091178

Computational theory of V1 receptive fields and binocular vision
SATOH Shunji; SAKAGUCHI Yutaka; USUI Shiro
We propose a simple but novel mathematical framework dealing with binocular vision; using complex functions, {left image} + i{right image}. This framework expands the solutions of mathematical problems on visual functions into complex functions. In this article, we derive the complex function as a solution of the differential equation generating the Gaussian derivative model for V1 receptive fields. We will show that (1) spatial receptive fields having orientation selectivity, (2) binocular receptive fields of V1 simple neurons, and (3) the energy model of V1 complex neurons encoding binocu..., The Institute of Electronics, Information and Communication Engineers, 02 Mar. 2010, IEICE technical report. Neurocomputing, 109, 461, 391-396, Japanese, 0913-5685, 110008004332, AN10091178

一般化Gaussian DerivativeによるV1受容野のモデル—両眼視差・運動方向選択性受容野モデルとそ工学的利点—
佐藤俊治; 臼井支朗; 阪口豊
24 Sep. 2009, 日本神経回路学会全国大会講演論文集, 19th, 120-121, Japanese, 200902233469868822

Fractional derivative of Gaussian functions : A model for spatio-temporal receptive fields of V1 simple cells
SATOH Shunji; USUI Shiro
A novel model for spatio-temporal receptive fields of V1 simple neurons is proposed by introducing fractional order derivatives. We consider the computational role of V1 simple neurons, and derive physiologically acceptable and theoretically optimal image-bases representing retinal images. The image-bases are derived as solutions of a variational problem which formulates the necessities of visual processing by V1 neurons. Using the derived image-bases, a new mathematical model for spatio-temporal receptive fields is derived. A model for binocular V1 neurons selective to spatial disparities ..., The Institute of Electronics, Information and Communication Engineers, 04 Mar. 2009, IEICE technical report. Neurocomputing, 108, 480, 141-146, Japanese, 0913-5685, 110007324973, AN10091178

Platform for a collaborative brain system modeling
KANNON Takayuki; MAKIMURA Koji; KAMIJI Nilton Liuji; SATOH Shunji; USUII Shiro
Mathematical model of the brain developed so far have targeted specific brain areas such as, cells, networks and/or phenomena, including microscopic and macroscopic levels. However, in order to understand the whole brain system, integration of such specific models and large scale simulation studies are necessary. To tackle this issue, we are developing a collaborative platform implemented as an Eclipse plug-in. It consists of: the Concierge plug-in used to store, search and manage experimental data, paper PDFs, etc, which have being extended for the management of mathematical models as well..., The Institute of Electronics, Information and Communication Engineers, 04 Mar. 2009, IEICE technical report. Neurocomputing, 108, 480, 37-42, Japanese, 0913-5685, 110007325008, AN10091178

Computational model for filling-in at the blind spot utilizing regularization theory and biological findings
SATOH Shunji
A visual model for filling-in at the blind spot is proposed. Standard regularization theory is employed to derive a visual model deductively. First, some problems of the diffusion eqation, which is frequently used for depth perception, are pointed out. Then, the computational role of a neural property discovered by Matsumoto and Komatsu (J. Neurophysiology, vol.93, pp.2374-2387, 2005) is reconsidered, and second derivative quantities, curvature of level-set and curvature of flow line, are appended into a priori knowledge of image on the blind spot. Moreover, two different kinds of informati..., The Institute of Electronics, Information and Communication Engineers, 09 Mar. 2007, IEICE technical report. Neurocomputing, 106, 590, 1-6, Japanese, 0913-5685, 110006249052, AN10091178

視覚的注意が図地割り当てに及ぼす影響
日髙 聡太; 佐藤 俊治; 行場 次朗
公益社団法人 日本心理学会, 03 Nov. 2006, 日本心理学会大会発表論文集, 70, 3PM044-3PM044, Japanese, 2433-7609, 130007393327

A computational V1 model derived from multi-resolution image-reconstruction
SATOH Shunji
Neurophysiological experiments have revealed that post-synaptic effect conveyed by long-range horizontal connections (LHC) in V1 can be excitatory or inhibitory, and the effect highly depends on the activity of pre-synaptic neurons (non-linearity of horizontal connections). Almost all of computational studies about LHC have been based on contour integration. However, contour integration cannot explain the non-linearity of LHC. In this study, the author proposes image-reconstruction as the computational role of LHC in V1 instead of contour integration. The spatial distribution of LHC and the..., The Institute of Electronics, Information and Communication Engineers, 09 Mar. 2006, IEICE technical report. Neurocomputing, 105, 658, 55-60, English, 0913-5685, 110004680367, AN10091178

V1野短距離水平結合の計算原理
佐藤 俊治
2006, 日本神経回路学会全国大会

動的輪郭法を計算原理とする視覚モデル ―工学的画像処理手法は脳の数理モデルにもなり得る―
佐藤俊治
日本工業出版, Dec. 2005, 画像ラボ, 16, 12, 22-26, Japanese, Introduction other, 0915-6755, 40007054781, AN10164169

A Visual Model Based on the Computational Principle of Level-Set Method and Active Contours
SATOH Shunji
A visual model for object detection is proposed. The computational principles of active contours and level-set method are utilized to construct the proposed model. A visual model for object detection have been proposed by Satoh (IEICE Tech. Rep. Vol. HIP2004-63 (2004)), it has, however, problems : (i) long computational time, (ii) low noise tolerance for large images. In this report, a generalized derivative with spatial scale is introduced in order to tackle the problems. Numerical simulations show that the problems are partially solved, and the proposed model explain the relation between ..., The Institute of Electronics, Information and Communication Engineers, 23 Mar. 2005, IEICE technical report. Neurocomputing, 104, 760, 1-6, Japanese, 0913-5685, 110003234179, AN10091178

V1野における長距離水平結合の計算原理
佐藤 俊治
2005, 日本神経回路学会第15回全国大会講演論文集 1, 93

A Visual Model Based on the Computational Principle of Active Contours
SATOH Shunji
A visual model for object detection is proposed. The dynamics of the proposed model is based on the computational principle of active contours, which is a fundamental method for various image processing. A redefined level set function is utilized in order to model object-detecting neurons. An initial value problem of active contours is solved by introducing a visual property revealed by cognitive psychological experiments. Numerical experiments show that the technical effectiveness of the proposed model. The proposed model detects objects with complex topologies, which has not been dealt wi..., The Institute of Electronics, Information and Communication Engineers, 09 Dec. 2004, Technical report of IEICE. HIP, 104, 525, 1-6, Japanese, 0913-5685, 110003272576, AN10487237

H-003 A Study of Perceptual Rivalry by Energy Analysis of Active Contour
Satoh Shunji; Miyake Shogo
FIT(電子情報通信学会・情報処理学会)推進委員会, 20 Aug. 2004, 情報科学技術フォーラム一般講演論文集, 3, 2, 409-411, Japanese, 110007683879, AA11740605

Computational Model for Working Memory Controled by Reward System
SASAKI Ryo; MIYAKE Shogo; SATOH Shunji
A computational model for working memory which is controled by a reward system is proposed. The control system of working memory is formulated by use of a temporal difference model which estimates future reward. Through the computational experiment, the behavior of the model is studied for a delayed response task and and a Go-Nogo task. It is pointed out that the prediction of the other various attention signals is important as well as the information of reward., The Institute of Electronics, Information and Communication Engineers, 12 Mar. 2004, IEICE technical report. Neurocomputing, 103, 734, 85-90, Japanese, 0913-5685, 110003232593, AN10091178

A model of long-term memory based on cholinergic control
SHISHIDO Hideaki; MIYAKE Shogo; SATOH Shunji
From the recent physiological data, it is shown that acetylcholine controls the synaptic transmission in the hippocampus and induces the theta rhythm and the rhythm of sleep and waking. It is also suggested that acetylcholine plays a very important role in the formation of long-term memory. We proposed a hippocampus-cortex model for long-term memory with effects of acetylcholine, theta rhythm and sleep. We investigated the process of the formation of long-term memory by computer simulations and clarified the roles of acetylcholine and theta rhythm in the process., The Institute of Electronics, Information and Communication Engineers, 12 Mar. 2004, IEICE technical report. Neurocomputing, 103, 734, 91-96, Japanese, 0913-5685, 110003232594, AN10091178

The Knowledge Acquisition Method by Neural Network Using Nonlinear Transformation
NAKAMURA Ikuo; SATOH Shunji; MIYAKE Shogo; ASO Hirotomo
Multilayer perceptron is widely used as high-performance discrimination circuit. Learned knowledge is expressed as connection weights and output values in hidden and output layer. However it is difficult to gain the comprehensible knowledge from the trained network because of its complicated connections and nonlinear transformations. In this paper, we propose a extraction method of polynomial rules. It trains characteristic traits of each input elements with two or more nonlinear functions prepared for the input part of multilayer perceptron. Moreover both parameters and weights of these fu..., The Institute of Electronics, Information and Communication Engineers, 11 Mar. 2004, IEICE technical report. Neurocomputing, 103, 733, 61-66, Japanese, 0913-5685, 110003232560, AN10091178

Visual Attention Model in Figure-Ground Reversal Perception
OHIZUMI Yoji; SATOH Shunji; MIYAKE Shogo; ASO Hirotomo
In figure-ground reversal perception for ambiguous figures, figure-ground relationship reverses depending on the position and the range of our attention. Our attention also changes depending on figural features. We propose a neural network model that explains a dynamic process concerning figure-ground perception and visual attention. Our model consists of two mechanisms: (i) one determines figure-ground relationship between regions by use of contour information, and (ii) another gives influence of visual attention on (i). These two mechanisms affect each other according to the states of fig..., The Institute of Electronics, Information and Communication Engineers, 10 Mar. 2004, IEICE technical report. Neurocomputing, 103, 732, 117-122, Japanese, 0913-5685, 110003232540, AN10091178

A-15-21 A Visual Model of Boundary Detection using Lateral Connections
Kamiyama Yutaka; Satoh Shunji; Miyake Shogo
The Institute of Electronics, Information and Communication Engineers, 08 Mar. 2004, Proceedings of the IEICE General Conference, 2004, 301-301, Japanese, 110003265940, AN10471452

Pattern Recognition Model with Moving Visual Attention Based on Hypothesis and Verification
SHIMOMURA Masao; SATOH Shunji; MIYAKE Syogo; ASO Hirotomo
認識開始時において網膜上にパターン全体が射影されていない場合でも,注視点を動かすことによりパターンを正しく切出し・認識可能なモデルを提案する.本モデルは,まず提示位置に影響されない特徴を用いてパターンの仮の認識結果(仮定)を生成し,続いて提示位置に依存する特徴を用いて,得られた仮定を検証しながら注視点の移動を行うという動作を繰り返すことにより,適切なパターンの切出し・認識を実現する.また,本モデルの計算機シミュレーションを行うことで,実際にパターン一部分の提示から適切な切出し・認識が可能であることを示す.更に,複数のパターンが混在して提示されている状況やノイズが混入した状況から認識を開始しても適切に認識動作が行えることを示す., The Institute of Electronics, Information and Communication Engineers, 01 Aug. 2003, The transactions of the Institute of Electronics, Information and Communication Engineers. D-II, 86, 8, 1244-1253, Japanese, 0915-1923, 110003170963, AA11340957

Pattern Recognition Model with Moving Visual Attention Based on Hypothesis and Verification
SHIMOMURA Masao; SATOH Shunji; MIYAKE Syogo; ASO Hirotomo
The Institute of Electronics, Information and Communication Engineers, 01 Aug. 2003, IEICE transactions on information and systems, 86, 8, 1475-1476, English, 0916-8532, 110003223343, AA10826272

A Model for a Radial-Arms Maze Task using Reinforcement Learning
Yamazaki Susumu; Ito Makoto; Miyake Shogo; Satoh Shunji
We construct a model of a radial-arms maze task for rats in which a higher function of memory is required. We formulate a radial-arms maze task as a reinforcement learning problem. In this model, agent can successfully learn depnding on past memory by adding feedback to an Actor-Critic method that is one of reinforcement learning algorithms. Simulations show that the model explains a result of the behavioral experiment of a rat well., The Institute of Electronics, Information and Communication Engineers, 12 Mar. 2003, IEICE technical report. Neurocomputing, 102, 731, 125-130, Japanese, 0913-5685, 110003232372, AN10091178

A model of visual attention with Gestalt principles on shapes of objects
SATOH Shunji; MIYAKE Shogo
Principles of visual grouping and figure-ground relationship about contours of objects have been identified as closure, symmetry, convexity, parallelism, proximity and so on. These principles should be considered in order to construct an abject recognition model for any objects including unknown objects. However, there is no need to formulate every principle because these priniciples are descriptive and correlate with each other. Therefore, we propose the generalized symmetry transform. The generalized symmetry transform is employed by the visual attention model proposed by authors in order..., The Institute of Electronics, Information and Communication Engineers, 11 Mar. 2003, IEICE technical report. Neurocomputing, 102, 730, 79-84, Japanese, 0913-5685, 110003232334, AN10091178

The long-term prediction for non-stationary time serieses
Kimura Hiroshi; Satoh Shunji; Sugaya Yoshihiro; Miyake Shogo; Aso Hirotomo
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 61-61, Japanese, 110003238941, AN10471452

A self-organizing network for of learning Markov processes
Fukui Masaaki; Miyake Shogo; Satoh Shunji
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 1, 18-18, Japanese, 110003239312, AN10471452

A neural network model for figure detection by use of pulse-coupled neurons
Satoh Yoshihiro; Satoh Shunji; Miyake Shogo; Aso Hirotomo
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 1, 21-21, Japanese, 110003239315, AN10471452

A learning model for sequential motor tasks using hierarchical reinforcement learning
Sasaki Yoshinori; Satoh Shunji; Miyake Shogo
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 1, 25-25, Japanese, 110003239319, AN10471452

A model for perception and generation of gestures based on motor theory of perception
Imazawa Yoshiro; Satoh Shunji; Miyake Shogo
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 1, 26-26, Japanese, 110003239320, AN10471452

Kernel Based Adaptive Dimensionality Reduction for Pattern Recognition
Sasaki Yuji; Kato Tsuyoshi; Satoh Shunji; Miyake Shogo; Aso Hirotomo
The Institute of Electronics, Information and Communication Engineers, 03 Mar. 2003, Proceedings of the IEICE General Conference, 2003, 2, 204-204, Japanese, 110003239766, AN10471452

A model of visual attention based on scale-space analysis and cognitive phenomena of humans
SATOH Shunji; MIYAKE Shogo
A model for visual attention based on scale-space theory is proposed by employing results given by neurophysiological experiments and cognitive phenomena of humans on the vision. The major parts of the model are (a) a model of early vision, and (b) a computational model for focus of attention (FOA) . First, it is indicated that the model of early vision can be described as a discretized scale-space, and equations denoting the behavior of early vision are given. Next, the FOA model is constructed based on scale-space theory. Numerical simulations show that (1) the model successively gives at..., The Institute of Electronics, Information and Communication Engineers, 27 Jan. 2003, IEICE technical report. Neurocomputing, 102, 627, 37-42, Japanese, 0913-5685, 110003232257, AN10091178

Utilization of a Modular-Type Neural Network with Nonlinear Transformation for Pattern Recognition Problems
MORISHIMA K; SATOH S; SHIMOMURA M; MIYAKE S; ASO H
A modular-type neural network is proposed in order to solve pattern recognition problems. Neurons in the hidden layer convert an input space to a curved space by use of polynominal functions, and a neuron in the output makes discriminant surfaces by RBF(Radial Basis Functions). The proposed network is able to recognize patterns whose distribution of the features are non-super-ellipsoids. We show that the network has ability to make a multiple discriminant surface, that is, multi-templates are generated automatically. Numeral simulations show that the network works successfully for different..., The Institute of Electronics, Information and Communication Engineers, 12 Mar. 2002, IEICE technical report. Neurocomputing, 101, 736, 95-102, Japanese, 0913-5685, 110003234291, AN10091178

Recognition of Occluded Patterns and Texture Patterns
Satoh S; Miyake S; Aso H; Takanaka K
A new recognition model for occluded patterns and texture patterns is proposed. The model includes two types of cells; cells that detect the luminance of input patterns and ones which have Gabor-like receptive fields. Generally, features of unknown patterns, ground patterns and trained patterns can not be estimated before recognition. The model groups a pattern by use of features in a pre-decided small region by introducing a function that makes the region. Grouped patterns are recognized by rotation-invariant neocognitron. A pattern which is not recognized by the recognition model is regar..., The Institute of Electronics, Information and Communication Engineers, 14 Mar. 2001, IEICE technical report. Neurocomputing, 100, 686, 55-62, Japanese, 0913-5685, 110003233814, AN10091178

Pattern Extraction Model Based on Hypothesis and Verification
Shimomura M; Satoh S; Miyake S; Aso H
We present a new neural network model for pattern extraction which is able to efficiently recognize patterns allocated in a plane. The model consists of two units, a saccade unit with low-resolution and a gazing unit with high-resolution. Both units have fundamentally same architecture in which a neocognitron works out features of image;the hypothesis is made based on shift-invariant features;then it is verified by use of characteristic positions of features. The verification process corrects a displacement of the pattern and tentavive result. Patterns are extracted during an iterative cycl..., The Institute of Electronics, Information and Communication Engineers, 14 Mar. 2001, IEICE technical report. Neurocomputing, 100, 686, 63-70, Japanese, 0913-5685, 110003233815, AN10091178

A feature extraction method using independent component analysis.
SUDOH Takashi; SATOH Shunji; MIYAKE Shogo; ASO Hirotomo
The method of feature extraction in usual pattern recognition depends heavily on objects and is heuristic. In order to get high performance in pattern recognition, feature extraction is an important process. There are only few studies which discuss on general feature extraction for any object. In this paper, we study a general feature extraction method for any object. We use independent conponent analysis for it. We examine the method for handwritten digit and texture images which have very different attributes each other, and evaluate the efficiency of the extracted features. Further, we d..., The Institute of Electronics, Information and Communication Engineers, 14 Mar. 2001, IEICE technical report. Neurocomputing, 100, 686, 109-116, Japanese, 0913-5685, 110003233821, AN10091178

階層的パターン統合処理に基づく視覚情報処理モデル
佐藤 俊治
本論文は神経生理学や心理学的知見を導入して, パターンの多様性に影響されない視覚神経回路網モデルの構築を目的としており, 6章より構成される.第1章の「序論」に続いて, 第2章では, 既に提案されている視覚神経回路網モデルであるネオコグニトロンの構成・学習方法について定式化するとともに, 回転したパターンに頑健性がないことを実験により確認した.第3章では, 新しいボトムアップ型神経回路モデル(回転対応型ネオコグニトロン)を提案している.実際に手書き数字を用いた数値実刑により, パターンの変形・位置ずれ・拡大縮小・ノイズだけでなく, パターンの回転にも完全に頑健であることを示した.第4章では, 回転対応型ネオコグニトロンを含むネオコグニトロン型神経回路モデルの学習過程を解析し, その結果から高速に学習を行なうアルゴリズムを提案している.本アルゴリズムを用いることで, 認識性能に影響を及ぼすことなく, 学習時間が約1/680に短縮することを確認している.第5章では, 回転した文字を必ずしも瞬時に認識せず心的回転により初めて認識するというヒトの認識機能を実現する視覚モデルを提案している.数値実験により, パターンの多様性に頑健であることを明らかにした.また, 鏡像回転パターンに対する提案モデルの挙動が心理学的事実と符号するという興味深い結果も得られた.第6章「結論」では本論文の成果..., 社団法人人工知能学会, 01 Nov. 2000, 人工知能学会誌, 15, 6, 1004-1004, Japanese, 0912-8085, 110002808378, AN10067140

階層的パターン統合処理に基づく視覚情報処理モデル
佐藤 俊治
本論文は神経生理学や心理学的知見を導入して, パターンの多様性に影響されない視覚神経回路網モデルの構築を目的としており, 6章より構成される.第1章の「序論」に続いて, 第2章では, 既に提案されている視覚神経回路網モデルであるネオコグニトロンの構成・学習方法について定式化するとともに, 回転したパターンに頑健性がないことを実験により確認した.第3章では, 新しいボトムアップ型神経回路モデル(回転対応型ネオコグニトロン)を提案している.実際に手書き数字を用いた数値実刑により, パターンの変形・位置ずれ・拡大縮小・ノイズだけでなく, パターンの回転にも完全に頑健であることを示した.第4章では, 回転対応型ネオコグニトロンを含むネオコグニトロン型神経回路モデルの学習過程を解析し, その結果から高速に学習を行なうアルゴリズムを提案している.本アルゴリズムを用いることで, 認識性能に影響を及ぼすことなく, 学習時間が約1/680に短縮することを確認している.第5章では, 回転した文字を必ずしも瞬時に認識せず心的回転により初めて認識するというヒトの認識機能を実現する視覚モデルを提案している.数値実験により, パターンの多様性に頑健であることを明らかにした.また, 鏡像回転パターンに対する提案モデルの挙動が心理学的事実と符号するという興味深い結果も得られた.第6章「結論」では本論文の成果をまとめ, 今後の課題を述べている., 一般社団法人 人工知能学会, 01 Nov. 2000, 人工知能, 15, 6, 1004-1004, Japanese, 2188-2266, 2435-8614, 110002808378, AN10067140

Pattern Recognition Model Tolerant of the Position and Deformation Based on Hypothesis and Verification
SHIMOMURA Masao; SATOH Shunji; MIYAKE Shogo; ASO Hirotomo
The Institute of Electronics, Information and Communication Engineers, 07 Sep. 2000, Proceedings of the Society Conference of IEICE, 2000, 9-9, Japanese, 110003349282, AN10489017

Design and Imprementation of the Class Library for Construction of Neocognitron-type Neural Network Model
Miyano Y; Satoh S; Aso H; Miyake S
Neocognitron is a neural network model which can recognized noised, distorted and shifted patterns, and several extended models based on the neocognitron are proposed. However, the structures of these models become more and more complex in the process of the development for the excellent performance of the model, so that the much efforts are enforced in the development of the programming and in its test. In this paper, we propose a program library in C_<++> language to support construction neocognitron-type neural network models. The structure of this library coincides with the multi-layer ..., The Institute of Electronics, Information and Communication Engineers, 14 Mar. 2000, IEICE technical report. Neurocomputing, 99, 685, 137-144, Japanese, 0913-5685, 110003233658, AN10091178

Pattern Recognition System Based on Mental Rotation
Satoh S; Aso H; Miyake S; Kuroiwa J
A new neural network model which can recognize rotated, distorted, scaled, shifted and noised patterns is proposed. The model is constructed based on psychological experiments in a mental rotation. The model has two types of processes. One is a feed forward process in which pattern recognition is realized by means of a hybrid neocognitron, or a composition of a rotation-invariant neocognitron and a standard neocognitron. The other is a feedback process in which a mental rotation is executed by use of a model of associative recall in visual pattern recognition. The feedforward process produc..., The Institute of Electronics, Information and Communication Engineers, 19 Mar. 1999, IEICE technical report. Neurocomputing, 98, 674, 231-238, Japanese, 110003233535, AN10091178

Rotation invariant Neocognitron with plastical connections corresponding to rotational angle.
SATOH Shunji; KUROIWA Jousuke; ASO Hirotomo; MIYAKE Shogo
回転対応型ネオコグニトロンは, Fukushimaによって提案されたネオコグニトロンを, 回転したパターンの認識を可能とするように拡張した階層型神経回路モデルである. 我々は, 任意の回転パターンを認識可能とするために, 閾値制御法を用いた学習法を提案し, 最大Θ=120゜の認識が可能であることを数値シミュレーションで示した (Θは学習パターンからの角度ずれを示している). 本研究では任意の角度に回転させたパターン (例えばΘ=180゜) の認識を可能とするために, S層内の同一細胞面群内で回転角に対応した可変結合を持たせる手法を提案し, その有効性を確認する., The Institute of Electronics, Information and Communication Engineers, 06 Mar. 1997, Proceedings of the IEICE General Conference, 1997, 1, 25-25, Japanese, 110003260857, AN10471452

The improvment of rotation-invariant-Neocognitron
SATOH Shunji; KUROIWA Jousuke; ASO Hirotomo; MIYAKE Shogo
回転対応型ネオコグニトロンは,Fukushimaによって提案されたネオコグニトロンを拡張し,回転したパターンの認識をも可能にした階層型神経回路モデルである.回転した特徴を効率よく学習するために動的閾値制御法が提案されているが,本論文では学習パターン数に依存しない動的閾値制御法を提案する.また回転パターンの認識率と計算時間を更に向上させるため,細胞面群の生成と消滅を考慮した新しい学習方法を提案する., The Institute of Electronics, Information and Communication Engineers, 18 Sep. 1996, Proceedings of the Society Conference of IEICE, 1996, 21-21, Japanese, 110003336321, AN10489017

Neocognitron for recognition of rotated patterns
Satoh S; Kuroiwa J; Aso H; Miyake S; Inawashiro S
The neocognitron is a hierarchical neural network to recognize patterns after self organizing, and it is robust for deformation and/or shifting of patterns for recognition. However, it is difficult to recognize greatly rotated patterns. We propose a neocognitron that can recognize rotated patterns. By numerical simulation, it is found that the model can recognize rotated patterns without leaning the patterns themselves. Simply rotated patterns are regarded as entirly different patterns in conventional neocognitron, but new model can integrate both patterns and their rotated ones while learn..., The Institute of Electronics, Information and Communication Engineers, 18 Mar. 1996, IEICE technical report. Neurocomputing, 95, 598, 263-270, Japanese, 110003233140, AN10091178

偉人たちの健康診断
General book, Japanese, Supervisor, 幻冬舎, 01 Dec. 2019

総合コミュニケーション科学シリーズ ユニーク＆エキサイティング サイエンス
梶谷誠; 佐藤俊治; 崎山一男; 芳原容英; 椿美智子; 桂川眞行
Japanese, Joint work, 第1章 視覚の数学とプログラミング, 近代科学社, Apr. 2013

Knowledge-based Intelligent Techniques in Character Recognition
Shunji Satoh
Joint work, Chapter 3, CRC Press, 1999

Knowledge-based intelligent techniques in character recognition
English, Joint work, section3: Recognition of rotated patterns using a neocognitron, CRC Press, New York, 1999

錯視的色知覚の臨界融合周波数
新倉大輔; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会2021年冬季大会, 日本視覚学会, Domestic conference
22 Jan. 2021

視覚数理モデルシミュレーションの高速化と錯視画像の探索
柳田悠介; 佐藤俊治; 策力木格; 吉永努
Oral presentation, Japanese, 電子情報通信学ニューロコンピューティング研究会, Domestic conference
04 Mar. 2020

不同視状態における奥行き運動知覚特性の測定
上村浩平; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
11 Jan. 2020

自己運動中の移動物体知覚特性の計測と計算論的考察
成田侑毅; 佐藤俊治
Oral presentation, Japanese, 電子情報通信学会，ヒューマン情報処理研究会, Domestic conference
31 Oct. 2019

Auditory facilitation on visual detection task in the entire visual field
Yoshiyuki Sato; Takayuki Watanabe; Shunji Satoh
Poster presentation, English, Symposium of Yotta Informatics, International conference
20 Mar. 2019

サイティング時の優位眼偏心度依存性と両眼像融合
鎌田峻輔; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
31 Jan. 2019

錯視的知覚色の臨界融合周波数
鈴木悠介; 佐藤俊治; 中嶋 豊
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
31 Jan. 2019

長さ・位置・角度知覚の偏心度依存性
出水花織; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
31 Jan. 2019

両眼視差と運動視差が拡張現実における立体映像の位置知覚に与える効果
弓倉和恵; 佐藤俊治; 中嶋 豊
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
31 Jan. 2019

Virtual Reality 装置を用いた移動物体知覚特性の計測と計算論的考察
成田侑毅; 赤澤文彦; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
30 Jan. 2019

Virtual Reality 装置を用いた視覚実験のための調査と開発
赤澤文彦; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会冬季大会, Domestic conference
29 Jan. 2019

聴覚刺激による視覚刺激検出促進効果の水平視野全体における性質
佐藤好幸; 渡辺貴行; 佐藤俊治
Oral presentation, Japanese, 第10回多感覚研究会, Domestic conference
20 Oct. 2018

拡張現実映像に対する奥行き知覚特性
中嶋豊; 菊池雅大; 佐藤俊治
Oral presentation, Japanese, 日本心理学会第82回大会, Domestic conference
25 Sep. 2018

中低次の視覚計算論と錯視
Oral presentation, Japanese, 第65回脳科学ライフサポート研究センターセミナー, Domestic conference
22 Jun. 2018

画像工学的手段を用いた知覚体制化モデルの構築
田島有芸人; 佐藤俊治
Oral presentation, Japanese, 電子情報通信学会総合大会, Domestic conference
20 Mar. 2018

水平視差推定及び視差不定領域検出アルゴリズムの構築
米田浩貴; 佐藤俊治
Oral presentation, Japanese, 電子情報通信学会総合大会, Domestic conference
20 Mar. 2018

ゲインコントロールを実行する視覚数理モデルと錯視
佐藤俊治; 志賀亮紀
Oral presentation, Japanese, 視覚認識機能のモデル実現のための協調的システムの研究, Invited, 東北大学電気通信研究所, Domestic conference
02 Feb. 2018

視覚数理モデルによる錯視パターンの網羅的探索とその検証
中村大樹; 柳田悠介; 佐藤俊治; 吉永 努
Oral presentation, Japanese, 日本視覚学会2018年冬季大会, Domestic conference
18 Jan. 2018

境界帰属方向と物体の重なり順序を同一の計算理論により再現する視覚モデル
Zaem Arif Zainal; 佐藤俊治
Oral presentation, Japanese, 日本視覚学会2018年冬季大会, Domestic conference
18 Jan. 2018

第一次視覚野単純型細胞の受容野に関する数理モデル研究
上田一平; 佐藤俊治
Oral presentation, Japanese, 日本視覚学会2018年冬季大会, Domestic conference
18 Jan. 2018

コントラストゲインコントロールを実行する数理モデルと錯視
志賀亮紀; 佐藤俊治
Oral presentation, Japanese, 日本視覚学会2018年冬季大会, Domestic conference
17 Jan. 2018

Visiome data for computational study on the receptive fields of the primary visual cortex
Shunji Satoh; Ippei Ueda
Oral presentation, English, Advances in Neuroinformatics, International conference
20 Nov. 2017

Formulation of Border-Ownership Assignment in Area V2 as an Optimization Problem
Zaem Zainal; Shunji Satoh
Oral presentation, English, The 24th International Conference On Neural Information Processing (ICONIP), International conference
14 Nov. 2017

Gabor関数 vs. Gaussian微分関数；第1次視覚野単純型細胞の受容野モデルとしての比較
佐藤俊治
Poster presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
21 Sep. 2017

計算論的に最適な速度推定器よってMT野細胞の複雑な反応特性を説明す る
中村大樹; 佐藤俊治
Poster presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
20 Sep. 2017

回転中心軸動揺錯視に対する回転速度の影響
中嶋豊; 角田翔平; 佐藤俊治
Poster presentation, Japanese, 日本視覚学会夏季大会, Domestic conference
06 Sep. 2017

Auditory facilitation of visual speeded detection in the entire visual field
Yoshiyuki Sato; Takayuki Watanabe; Shunji Satoh
Oral presentation, English, European Conference on Visual Perception (ECVP), International conference
27 Aug. 2017

Online simulation environment for computational neuroscience and data analysis
Hidetoshi Ikeno; Tadashi Yamazaki; Takayuki Kannon; Yoshihiro Okumura; Yoshimi Kamiyama; Akito Ishihara; Keiichiro Inagaki; Yutaka Hirata; Shunji Satoh; Hiroaki Wagatsuma; Yoshiyuki Asai; Yoko Yamaguchi; Shiro Usui
Oral presentation, English, Neuroinformatics 2017, International conference
20 Aug. 2017

周辺視野における視覚情報処理に聴覚刺激が及ぼす影響
渡部貴行; 佐藤好幸; 佐藤俊治
Oral presentation, Japanese, 電子情報通信学会ヒューマン情報処理研究会, Domestic conference
09 Mar. 2017

曲面知覚の観察距離依存性に関する研究
青島初帆; 佐藤俊治
Oral presentation, Japanese, 映像情報メディア学会, Domestic conference
07 Mar. 2017

協調的視覚研究のための眼球モデルと網膜像計算
佐藤俊治; 出水花織
Invited oral presentation, Japanese, 東北大学共同プロジェクト研究会, Invited, Domestic conference
04 Feb. 2017

中低次の視覚計算問題の定式化とモデル比較
Zaem Zainal; 佐藤俊治
Invited oral presentation, Japanese, 東北大学共同プロジェクト研究会, Invited, Domestic conference
04 Feb. 2017

主観的輪郭で構成される図形の回転中心軸の知覚的動揺
中嶋豊; 角田翔平; 佐藤俊治
Poster presentation, Japanese, 日本基礎心理学会第35回大会, Domestic conference
29 Oct. 2016

Visiome Platform
Shunji Satoh
Poster presentation, English, Advances in Neuroinformatics, International conference
27 Nov. 2015

回転振動錯視に対する心理物理実験と計算論的考察
菊池勇作; 佐藤俊治
Oral presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
02 Sep. 2015

視覚研究用シミュレーション基盤: 数理モデルの結合，追加及び置換を行うための手法
占部一輝; 佐藤俊治; 中村大樹
Oral presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
02 Sep. 2015

簡素なMT 細胞モデルによる複雑な細胞特性の再現
中村大樹; 佐藤俊治
Oral presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
02 Sep. 2015

多義的な奥行知覚に関する心理物理実験と計算論的考察
満倉英一; 佐藤俊治
Oral presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
02 Sep. 2015

Solving problems in low/middle-order visual processing by applying a theorem of electrodynamics
Zaem Arif Zainal; Shunji Satoh
Oral presentation, Japanese, 日本神経回路学会全国大会, Domestic conference
02 Sep. 2015

Beyond the Gabor function for a receptive field model of V1 simple cells
Shunji Satoh
Invited oral presentation, English, Neuro 2015, Invited, International conference
30 Jul. 2015

協調的視覚モデル研究を目的としたRTミドルウェアの応用
佐藤俊治
Poster presentation, Japanese, 東北大学 電気通信研究所 共同プロジェクト研究発表会, Domestic conference
23 Feb. 2015

Neuroinformatics 的観点から構築された新規 Saliency map モデル
韓雪花; 佐藤俊治; 中村大樹; 占部一輝
Oral presentation, Japanese, 日本視覚学会, Domestic conference
22 Jan. 2015

視覚脳科学研究を目的としたRTミドルウェアの応用と結果
中村大樹; 佐藤俊治; 韓雪花; 占部一輝
Oral presentation, Japanese, 計測自動制御学会システムインテグレーション部門, Domestic conference
15 Dec. 2014

Visiome Platform: A Comprehensive Database for Vision Research
Shunji SATOH
Poster presentation, English, International Workshop of Advances in Neuroinformatics, International conference
25 Sep. 2014

Simulation Platform: Application Server for Testing and Sharing Mathematical Model and Experimental Data
Hidetoshi IKENO; Yoshimi KAMIYAMA; Akito ISHIHARA; Yutaka HIRATA; Shunji SATOH
Poster presentation, English, International Workshop of Advances in Neuroinformatics, International conference
25 Sep. 2014

MT細胞の電気生理実験結果に関する計算論的再考察
中村大樹; 佐藤俊治
Oral presentation, Japanese, 電子情報通信学会ニューロコンピューティング研究会, Domestic conference
17 Mar. 2014

複素関数による両眼性細胞の数理モデル化と画像処理への応用
広瀬正人; 佐藤俊治
Oral presentation, Japanese, 日本視覚学会2014年冬季大会
22 Jan. 2014

RT ミドルウエアをベースとした視覚研究用プラットフォームの開発
皆川保裕; 占部一輝; 佐藤俊治; 知久健; 川口仁; 長瀬雅之
Oral presentation, Japanese, 計測自動制御学会,第14回 計測自動制御学会 システムインテグレーション部門講演
Dec. 2013

視覚の数理モデル研究
佐藤俊治
Others, Japanese, 東京工業高等専門学校
Dec. 2013

Visiome-PF 委員会報告
佐藤俊治
Others, Japanese, NIJC運営会議・Platform 運用会議
Aug. 2013

運動知覚の計算理論ならびに数理モデル作成技術
佐藤俊治
Invited oral presentation, Japanese, 平成25年度第2回ブレインウェア研究会, 東北大学
Jun. 2013

速度知覚のパターン依存症に関する計算論的考察
飯野希; 中村大樹; 佐藤俊治
Oral presentation, Japanese, Vision,日本視覚学会２０１３年冬季大会
Jan. 2013

視覚運動情報の計算過程を記述する画像処理モデル
二枚田匠; 佐藤俊治
Oral presentation, Japanese, Vision,日本視覚学会２０１３年冬季大会
Jan. 2013

視覚数理モデル構築のためのプラットフォーム開発
北川大平; 占部一輝; 佐藤俊治
Oral presentation, Japanese, Vision,日本視覚学会２０１３年冬季大会
Jan. 2013

視覚を数理的に理解して画像処理アルゴリズムを作る
佐藤俊治
Invited oral presentation, Japanese, 情報処理学会北陸支部研究講演会, 情報処理学会
Nov. 2012

Visiome Platform
Yoshimi Kamiyama; Shin'ya Nishida; Shigeki Nakauchi; Izumi Ohzawa; Masao Tachibana; Takao Sato; Akiyoshi Kitaoka; Hiroshi Ashida; Hayaru Shouno; Shunji Satoh; Kazushi Maruya; Takayuki Kannon; Manabu Tanifuji; Shiro Usui
Oral presentation, English, Symposium of INCF Japan Node
Oct. 2012

錯視は「間違い」なのか？「ある意味正しい」のか？
佐藤俊治
Invited oral presentation, Japanese, 第5回錯覚ワークショップ, 第5回錯覚ワークショップ
Sep. 2012

Engineering and scientific approaches on vision science to develop novel algorithm and to solve paradox between physiology and perception.
Shunji Satoh
Invited oral presentation, English, 54th Mathematical Sciences based on Modeling, Analysis and Simulation seminar, Mathematical Sciences based on Modeling, Analysis and Simulation seminar, Kawasaki, Japan, International conference
Jul. 2012

協調的視覚研究の一考察
占部一輝; 佐藤俊治
Invited oral presentation, Japanese, 第3回視覚認識機能のモデル実現のための協調的システム, 東北大学
Mar. 2012

MT細胞特性と速度知覚特性の矛盾を計算論的に解く
佐藤俊治
Invited oral presentation, Japanese, 第3回視覚認識機能のモデル実現のための協調的システム, 東北大学
Mar. 2012

運動知覚特性の画像工学的解釈と計算論モデルに関する研究
外山敬介; 佐藤俊治
Oral presentation, Japanese, Vision
Jan. 2012

画像処理マシンとして視覚を理解し応用する
佐藤俊治
Invited oral presentation, Japanese, 豊田中央研究所
Sep. 2011

視覚神経系数理モデルシミュレーションのMPI による並列化
齋藤祐典; 佐藤俊治; 大村純一; 三好健文; 入江英嗣; 吉永努
Oral presentation, Japanese, ハイパフォーマンスコンピューティング研究発表会,ハイパフォーマンスコンピューティング研究会
May 2011

速度知覚に関する計算論的考察と心理物理実験
中畑達雄; 佐藤俊治; 阪口豊; 佐藤好幸
Oral presentation, Japanese, Vision,日本視覚学会2011年冬季大会
Jan. 2011

Brain science for image engineering and Engineers' view for vision science
Shunji Satoh
Invited oral presentation, English, 17th International Conference on Neural Information Processing, APNNA, Sydney, International conference
Nov. 2010

神経生理学的・工学的制約を考慮した視覚計算論研究
佐藤俊治
Oral presentation, Japanese, 神経科学・リハビリテーション・ロボット工学のシナジー効果に関する研究会,神経科学・リハビリテーション・ロボット工学のシナジー効果に関する研究会
Aug. 2010

An edge detection model reproducing nonlinear properties of V1
佐々木博昭; 佐藤俊治; 臼井支朗
Oral presentation, Japanese, システム制御情報学会,第54回システム制御情報学会研究発表講演会
May 2010

第一次視覚野における皮質内水平結合の計算論的意義
佐々木博昭; 佐藤俊治; 臼井支朗
Oral presentation, Japanese, 信学技報,ニューロコンピューティング研究会
2010

V1野細胞の受容野と視差選択性細胞に関する計算論的考察
佐藤俊治; 阪口豊; 臼井支朗
Oral presentation, Japanese, 信学技報,ニューロコンピューティング研究会
2010

初期視覚細胞による画像表現～画像工学・理論・神経生理学的観点からの考察と評価～
佐藤俊治
Public symposium, Japanese, ISシンポジウム, 電気通信大学大学院情報システム学研究科
Nov. 2009

A Next Generation Modeling Environment PLATO: Platform for Collaborative Brain System Modeling
Shiro Usui; Keiichiro Inagaki; Takayuki Kannon; Yoshimi Kamiyama; Shunji Satoh; Nilton L. Kamiji; Yutaka Hirata; Akito Ishihara; Hayaru Shouno
Invited oral presentation, English, International Conference on Neural Information Processing, APNNA, Bangkok, International conference
Oct. 2009

「視覚＝画像処理アルゴリズム」とみなして視覚を理解・応用する
佐藤俊治
Invited oral presentation, Japanese, BrainIS研究会, 九州工業大学
Jul. 2009

生理実験結果の計算論的解釈と工学的応用－受容野モデルと盲点補完を例として－
佐藤俊治
Invited oral presentation, Japanese, 視知覚研究の融合を目指して － 生理・心理物理・計算論, 生理学研究所
Jun. 2009

一般化Gaussian Derivative によるV1受容野のモデル～両眼視差・運動方向選択性受容野モデルとその工学的利点～
佐藤俊治; 臼井支朗; 阪口豊
Oral presentation, Japanese, 日本神経回路学会第19回全国大会,日本神経回路学会第19回全国大会
2009

V1野水平結合による符号化効率性の向上
佐々木博昭; 佐藤俊治; 臼井支朗
Oral presentation, Japanese, 日本神経回路学会第19回全国大会
2009

盲点補完の数理モデル―視覚研究で脳科学と画像工学へ同時に貢献したい―
佐藤俊治
Invited oral presentation, Japanese, 非線形動力学セミナー, 京都大学大学院
2007

視覚的注意が図地割り当てに及ぼす影響
日高聡太; 佐藤俊治; 行場次朗
Oral presentation, Japanese, 日本心理学会第70回大会
Nov. 2006

脳に学ばない脳のモデルで脳を理解し応用する
佐藤俊治
Invited oral presentation, Japanese, 日本知能ファジィ学会
2006

人間情報論１
電気通信大学

Advanced topics in perceptual system
The University of Electro-Communications

応用数学A
The University of Electro-Communications

知覚システム特論
The University of Electro-Communications

人間情報論1
The University of Electro-Communications

情報メディアシステム学基礎2
電気通信大学

Mechanical Engineering and Intelligent Systems Laboratory,Advanced Ⅱ
The University of Electro-Communications

Applied Mathematics A
The University of Electro-Communications

基礎演習A
The University of Electro-Communications

情報メディアシステム学基礎2
The University of Electro-Communications

Human informatics 1
The University of Electro-Communications

知能機械工学基礎実験第二
電気通信大学

Basic Exercises A
The University of Electro-Communications

基礎演習A
電気通信大学

人間情報論1
電気通信大学

応用数学A
電気通信大学

人間情報論1
The University of Electro-Communications

知覚システム特論
The University of Electro-Communications

知覚システム特論
電気通信大学

日本視覚学会

日本神経回路学会

電気通信情報学会

中次視覚機能の定式化と新しい機能的解釈による視覚計算理論
2016 - 2018

Visiomeプラットフォーム(PF)の継続開発・公開運用とコンテンツ収集・登録
2015 - 2016

眼光学シミュレータの開発
2016

脳と視覚に関する学術相談（秘密保持契約期間中）
2016

A computational study on early vision for scientific contribution and engineering contribution
SATOH Shunji
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, 独立行政法人理化学研究所->電気通信大学, Grant-in-Aid for Young Scientists (B), Principal investigator, Computational models of the early visual system were proposed based on theoretical analysis of spatio-temporal derivatives :(i) receptive fields of V1 neurons selective to binocular disparity, and(ii) selective to speed of moving stimuli. Moreover,(iii) a novel computational theory and a model of MT neurons were also derived from the theoretical analysis of the model(i) and(ii) above. As results of this MT research, complex neural properties of MT neurons and perceptual properties affected by image contrast were successfully accounted for by the single model of MT neurons. Software platform(s) have been developed to make computational research be more sustainable for large scale simulation of vision., 20700279
2008 - 2011

画像処理手法としても有効な統合的視覚モデルの研究
佐藤俊治
文部科学省, 科学研究費補助金(若手研究(B)), 東北福祉大学->独立行政法人理化学研究所, 若手研究(B), Principal investigator, 視覚野での情報処理を画像工学に応用するためには、線方位や曲率、面方位などの微分幾何的特徴量を計算する要素が必要となる。実際、17・18年度の当該研究で得られた種々の視覚モデルや理論は、画像の輝度勾配に対する1階微分(線方位検出)や2階微分(曲率など)の計算を必要としている。画像工学的にも多くの微分幾何量に基づく手法が提案されている。本年度の研究では、これら微分幾何量について神経生理学的観点から理論的に考察することで新しい受容野モデル(Weighted Hermite Function)を構築し、得られたモデルが神経生理学的実験結果を再現することを見出した。この結果は画像工学への貢献として、新しい画像フィルタの提案を意味する。具体的には以下の通りである。これまでのV1単純型細胞の受容野モデルとしては、Gabor filterやGaussian derivative model (GD)が採用されてきたが、それぞれには一長一短ある。例えばGabor filterは不確定性最小化の意味で最適画像フィルタであるが、特徴抽出の意味では最適ではない。GDは微分幾何との親和性が高いが、偶関数もしくは奇関数しか表現できず神経生理学的実験結果との対応が弱い。本研究ではまず、フーリエ空間に写像された画像特徴を効率よく抽出するための条件を考察した。この条件を満足する空間フィルタ(受容野)として、H..., 17700244
2005 - 2007

統合的視覚モデルの研究
佐藤俊治
文部科学省, 科学研究費補助金(若手研究(B)), 東北大学->東北福祉大学, 若手研究(B), Principal investigator, ヒトの視覚情報処理過程を説明する数理モデルは、視覚の柔軟性や汎化性を考慮すると、画像工学的にも有効である必要がある。そこで、物体検出を行なっているV4野細胞の動作原理として、動的輪郭法を採用し、物体検出を行なう視覚モデルを提案した。また、動的輪郭法が抱える問題を、認知心理学研究で得られた視覚特性(凸性)を導入することで解決した。神経生理学的実験により、視覚的注意の影響はV4細胞の活動度に影響を及ぼすことがわかっている。そこで、提案した物体検出モデルに視覚的注意の効果を導入した、統合的な視覚モデルの提案を行なった。統合的視覚モデルの妥当性を評価するために、図地反転現象に関する視覚心理実験を行なった。この心理実験結果と、モデルの動作特性が一致することを数値シミュレーションにより見出した。さらに、提案モデルの理論的な動作解析から、注意の範囲が知覚に影響を及ぼすことを予測した。この予測の妥当性を評価するために現在、新しい心理実験を行なっている。提案モデルは基本的に、反応拡散方程式に基づいて動作するが、結果を得るまでに長い時間を要するという問題点があった。そこでこの問題を解決するために、多解像度理論であるスケールスペース理論を用いて、スケールで一般化された微分演算子を統合的視覚モデルに導入した。数値シミュレーションにより上記問題点が解決され、さらに、V4野における長距離水平結合の計算論..., 14780254
2002 - 2004

表示装置、表示制御方法、及び表示制御プログラム
Patent right, 出願, Date applied: 2018

盲点補完を利用した画像空間フィルタを用いた画像制御方法
Patent right, 特願2008-086210, Date applied: 2008, 特開2009-239829

ニューロインフォマティクス・ニュースレター（理化学研究所発刊）に研究紹介記事が掲載される．
http://www.neuroinf.jp/modules/news/index.php?page=article&storyid=144&ml_lang=ja
2015 - 2015