Faculty Profiles - Nakazawa Atsushi

Corneal Reflection Analysis for Point of Gaze Estimation and Other Applications

Atsushi Nakazawa, Christian Nitschke

The 7th International Workshop on Robust Computer Vision 2013

Arm Pose Copying for Humanoid Robots

Yasser Mohanamad, Toyoaki Nishida, Atsushi Nakazawa

2013 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND BIOMIMETICS (ROBIO) 897 - 904 2013

Language：English Publisher：IEEE

Learning by imitation is becoming increasingly important for teaching humanoid robots new skills. The simplest form of imitation is behavior copying in which the robot is minimizing the difference between its perceived motion and that of the imitated agent. One problem that must be solved even in this simplest of all imitation tasks is calculating the learner's pose corresponding to the perceived pose of the agent it is imitating. This paper presents a general framework for solving this problem in closed form for the arms of a generalized humanoid robot of which most available humanoids are special cases. The paper also reports the evaluation of the proposed system for real and simulated robots.

DOI： 10.1109/ROBIO.2013.6739576

Point of Gaze Estimation through Corneal Surface Reflection in an Active Illumination Environment

Atsushi Nakazawa, Christian Nitschke

Image sensing symposium 2013

角膜イメージング法を用いた校正無し注視点推定

中澤篤志, 中澤篤志, ニチュケクリスティアン

日本ロボット学会学術講演会予稿集(CD-ROM) 31st 2013

I see what you see: Point of Gaze Estimation from Corneal Images

Christian Nitschke, Atsushi Nakazawa, Toyoaki Nishida

2013 SECOND IAPR ASIAN CONFERENCE ON PATTERN RECOGNITION (ACPR 2013) 298 - 304 2013

Language：English Publisher：IEEE COMPUTER SOC

Eye-gaze tracking (EGT) is an important problem with a long history and various applications. However, state-of-the-art geometric vision-based techniques still suffer from major limitations, especially (1) the requirement for calibration of a static relationship between eye camera and scene, and (2) a parallax error that occurs when the depth of the scene varies. This paper introduces a novel concept for EGT that overcomes these limitations using corneal imaging. Based on the observation that the cornea reflects the surrounding scene over a wide field of view, it is shown how to extract that information and determine the point of gaze (PoG) directly in an eye image. To realize this, a closed-form solution is developed to obtain the gaze-reflection point (GRP), where light from the PoG reflects at the corneal surface into a camera. This includes compensation for the individual offset between optical and visual axis. Quantitative and qualitative evaluation shows that the strategy achieves considerable accuracy and successfully supports depth-varying environments. The novel approach provides important practical advantages, including reduced intrusiveness and complexity, and support for flexible dynamic setups, non-planar scenes and outdoor application.

DOI： 10.1109/ACPR.2013.84

Virtual Dance Hall : Dance Interaction using Body Movement

YASUNAGA Takuya, NAKAZAWA Atsushi, TAKEMURA Haruo

Technical report of IEICE. Multimedia and virtual environment 112 ( 221 ) 61 - 66 2012.9

Language：Japanese Publisher：The Institute of Electronics, Information and Communication Engineers

This paper presents a new approach for synthesizing dance performance matched to input music, based on the emotional aspects of dance performance. Our analysis method extracts motion rhythm and intensity from motion capture data and musical rhythm, structure from musical signals. We extract candidates of motion segment sets whose features are matched to those of music segments. For synthesizing dance performance, we select the motion segment set whose intensity by Kinect and connectivity is matched to that of music segments. As the result, our system adds an interactive component to live dancing performed by virtual characters.

Super resolution scene reconstruction using corneal reflections

NITSCHKE Christian, NAKAZAWA Atsushi

Technical report of IEICE. PRMU 112 ( 225 ) 65 - 70 2012.9

Language：Japanese Publisher：The Institute of Electronics, Information and Communication Engineers

The corneal imaging technique enables extraction of scene information from corneal reflections and realizes a large number of applications including environment map reconstruction and estimation of a person's area of view. However, since corneal reflection images are usually low quality and resolution, the outcome of the technique is currently limited. To overcome this issue, we propose a first non-central catadioptric approach to reconstruct high-resolution scene information from a series of lower resolution corneal images through a super-resolution technique. We describe a three-step process, including (1) single image environment map recovery, (2) multiple image registration, and (3) high-resolution image reconstruction. In a number of experiments we show that the proposed strategy successfully recovers high-frequency textures that are lost in the source images, and also works with other non-central catadioptric systems, e.g., involving spherical mirrors.

Human-Computer Dance Interaction with Real-time Control using Wiimote and Kinect

YASUNAGA Takuya, NAKAZAWA Atsushi, TAKEMURA Haruo

Technical report of IEICE. PRMU 112 ( 225 ) 23 - 28 2012.9

Language：Japanese Publisher：The Institute of Electronics, Information and Communication Engineers

This paper presents a new approach for synthesizing dance performance matched to input music, based on the emotional aspects of dance performance. Since the sensor reads express intensities of users' body movements, the system can synthesize character motions whose intensities are synchronized to those of users. Our analysis method extracts motion rhythm and intensity from motion capture data and musical rhythm, structure from musical signals. With the results of this analysis, we generate a motion graph that can generate character motions matched to the musical rhythm. For synthesizing dance performance, we select the motion segment set whose intensity by sensor data and connectivity is matched to that of music segments. As the results, our system adds an interactive component to live dancing performed by virtual characters.

大阪大学における新入生の普通教科「情報」に関する調査報告

間下以大, 清川清, 中澤篤志, 竹村治雄

情報教育シンポジウム2012論文集 2012 ( 4 ) 29 - 34 2012.8

Educational Computer System in Osaka University

2012 ( 8 ) 1 - 5 2012.5

A Instrumented Puppet Interface for Retrieval of Motion Capture Data

NUMAGUCH Naoki, NAKAZAWA Atsushi, SHIRATORI Takaaki, HODGINS Jessica

Technical report of IEICE. PRMU 111 ( 430 ) 41 - 46 2012.2

Language：Japanese Publisher：The Institute of Electronics, Information and Communication Engineers

Intuitive and efficient retrieval of motion capture data is essential for effective use of motion capture databases. In this paper, we describe a system that allows the user to retrieve a particular sequence by performing an approximation of the motion with an instrumented puppet. This interface is intuitive because both adults and children have experience playacting with puppets. We develop a novel similarity metric called dual subspace projection method (DSPM) which works even for relatively large databases. We conducted a user study of the system and subjects could find the desired motion with reasonable accuracy from a database consisting of everyday, exercise, and acrobatic behaviors.

Dance to Music Character Animation with accelerometer based user control

Yasunaga Takuya, Nakazawa Atsushi, Takemura Haruo

IPSJ SIG Notes 2012 ( 25 ) 1 - 6 2012.1

Language：Japanese Publisher：Information Processing Society of Japan (IPSJ)

This paper presents a new approach for synthesizing dance performance matched to input music, based on the emotional aspects of dance performance. Our analysis method extracts motion rhythm and inten- sity from motion capture data and musical rhythm, structure from musical signals. We rst extract candidates of motion segment sets whose features are matched to those of music segments. For synthesizing dance perfor- mance, we select the motion segment set whose intensity by accelerometer and connectivity is matched to that of music segments. We evaluate all synthesizing time and matching value.

Human-computer dance interaction with realtime accelerometer control Reviewed

Takuya Yasunaga, Atsushi Nakazawa, Haruo Takemura

MM 2012 - Proceedings of the 20th ACM International Conference on Multimedia 1157 - 1160 2012

Language：English

Motion-capture-based character animations are widely used in computer graphics and interactive games.In this paper, we show a novel approach to create dancing character animations that react to input music and an accelerometer manipulated by a user. Since the sensor reads express intensities of users' body movements, the system can synthesize character motions whose intensities are synchronized to those of users. Our system consists of analysis phase and synthesis phase. In the analysis phase, the musical beat and segments are detected from input sound, and motion rhythm and intensities are found from motion capture data. With the results of this analysis, we generate a motion graph that can generate character motions matched to the musical rhythm. In synthesis phase, the system receives the output data from an accelerometer and traverses the motion graph according to the matching result between the sensor data and the motion intensity. As the result, our system adds an interactive component to live dancing performed by virtual characters. © 2012 ACM.

DOI： 10.1145/2393347.2396407

Scopus

大阪大学の情報教育システムに関する現状と展望

中澤篤志, 間下以大, 清川清

大学ICT推進協議会年次大会論文集 5p 2012

Super-Resolution from Corneal Images

Christian Nitschke, Atsushi Nakazawa

PROCEEDINGS OF THE BRITISH MACHINE VISION CONFERENCE 2012 2012

Language：English Publisher：B M V A PRESS

DOI： 10.5244/C.26.22

Virtual Dance Hall : Dance Interaction using Body Movement

14 61 - 66 2012

A Instrumented Puppet Interface for Retrieval of Motion Capture Data

2011 ( 1 ) 1 - 8 2011.11

眼球の表面反射と高速アクティブ光投影を用いた非装着・事前校正不要な注視点推定

中澤篤志, ニチュケクリスチャン, ラドコフアレクサンダー, 竹村治雄

画像の認識・理解シンポジウム(MIRU2011)論文集 2011 41 - 48 2011.7

Image-based Eye Pose and Reflection Analysis for Advanced Interaction Techniques and Scene Understanding

Christian Nitschke, Atsushi Nakazawa, Haruo Takemura

研究報告コンピュータビジョンとイメージメディア（CVIM） 2011 ( 31 ) 1 - 16 2011.5

Language：English

Recently, the geometric relation between a human eye and its image has been formalized to analyze environmental light reflections in the cornea. Proceeding with these efforts, our study proposes a theory of the light transport at the corneal surface including multiple eye poses, develops novel applications, and performs comprehensive experimental evaluation. Based on anthropometric data, a spherical-curvature geometric eye model is developed, and subsequently applied to discuss methods for eye pose estimation from projected circular eye features. The combination of camera and corneal mirror acts as a catadioptric imaging system, for which we describe the back projection to reconstruct the position of a light source, and the forward projection of light from a given source into the image. The theory has several practical applications in scene reconstruction and human-computer interaction, where we discuss the geometric calibration of display-camera setups as one particular problem. We propose a novel approach that eliminates the requirement of special hardware and tedious user interaction by analyzing screen reflections in the cornea. Based on this setting, thorough experimental evaluation shows that simple scene reconstruction results in a large error. We discuss possible reasons and introduce an optimization scheme that achieves feasible results by exploiting geometry constraints within the system. Our study provides sophisticated strategies for analyzing the geometric relation between camera, eye pose, corneal shape, and scene structure within arbitrary dynamic environments. The findings and developments enable novel insights, understanding, and applications in the analysis of human-scene interaction. We believe that this work has implications on several fields and is an important contribution.Recently, the geometric relation between a human eye and its image has been formalized to analyze environmental light reflections in the cornea. Proceeding with these efforts, our study proposes a theory of the light transport at the corneal surface including multiple eye poses, develops novel applications, and performs comprehensive experimental evaluation. Based on anthropometric data, a spherical-curvature geometric eye model is developed, and subsequently applied to discuss methods for eye pose estimation from projected circular eye features. The combination of camera and corneal mirror acts as a catadioptric imaging system, for which we describe the back projection to reconstruct the position of a light source, and the forward projection of light from a given source into the image. The theory has several practical applications in scene reconstruction and human-computer interaction, where we discuss the geometric calibration of display-camera setups as one particular problem. We propose a novel approach that eliminates the requirement of special hardware and tedious user interaction by analyzing screen reflections in the cornea. Based on this setting, thorough experimental evaluation shows that simple scene reconstruction results in a large error. We discuss possible reasons and introduce an optimization scheme that achieves feasible results by exploiting geometry constraints within the system. Our study provides sophisticated strategies for analyzing the geometric relation between camera, eye pose, corneal shape, and scene structure within arbitrary dynamic environments. The findings and developments enable novel insights, understanding, and applications in the analysis of human-scene interaction. We believe that this work has implications on several fields and is an important contribution.