Faculty Profiles - Nakazawa Atsushi

MISC - Nakazawa Atsushi

Division display 101 - 120 of about 169 ／ All the affair displays >>

Report on CVPR2008

KAWASAKI Hiroshi, SHIMIZU Masao, TAKAMATSU Jun, TANAKA Masayuki, NAKAZAWA Atsushi, NOBUHARA Shohei, FURUKAWA Ryo, LAO Shihong, YAGI Yasushi

IEICE technical report 108 ( 328 ) 145 - 152 2008.11

Report on CVPR2008

KAWASAKI Hiroshi, SHIMIZU Masao, TAKAMATSU Jun, TANAKA Masayuki, NAKAZAWA Atsushi, NOBUHARA Shohei, FURUKAWA Ryo, LAO Shihong, YAGI Yasushi

IPSJ SIG Notes. CVIM 2008 ( 115 ) 145 - 152 2008.11

Display-Camera Calibration from Eye Reflections

NITSCHKE Christian, NAKAZAWA Atsushi, TAKEMURA Haruo

Technical report of IEICE. PRMU 108 ( 198 ) 113 - 120 2008.8

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

We present a technique for calibrating display-camera systems from corneal reflections in the user's eyes. Display-camera systems enable a range of vision applications that need controlled illumination including 3D object reconstruction, facial modeling or human computer interaction in everyday environments. An important issue is calibrating the pose of the display with respect to the camera. Such a calibration may be achieved using a planar mirror with attached calibration pattern or a spherical mirror of known size. However, all approaches require additional hardware and user interaction. We propose an automatic way to recover display properties from patterns that are reflected in the cornea, a mirroring device that naturally coexists in any display-camera system. By applying this strategy we also obtain a continuous estimation of eye pose which may be used to calibrate eye tracking systems and generally enhance human-computer interaction.

CiNii Article

CiNii Books

researchmap

Human pose estimation using volumetric features and boosting approach

TAIRA Ryosuke, NAKAZAWA Atsushi, TAKEMURA Haruo

IPSJ SIG Notes. CVIM 2008 ( 82 ) 143 - 148 2008.8

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

Recently, considerable research has been conducted on markerless human-body motion capture using volume data. Most studies have used articulated body models that consist of primitives such as cylinders or ellipsoids. However, the methods that use such models require very good initial parameters. This paper proposes a method to find the human posture parameter from just a single frame without any knowledge of previous frames. We first prepare sample volume data which takes various postures, and then cluster them according to the direction of body links. The volumetric feature vector is acquired for each sample and they are learned via the AdaBoost algorithm. The resulting classifier is used for estimating input volume data, and the matched class can be used as an initial parameter for tracking based methods.

CiNii Article

CiNii Books

researchmap

Other Link： http://id.nii.ac.jp/1001/00051779/

Display-Camera Calibration from Eye Reflections

NITSCHKE Christian, NAKAZAWA Atsushi, TAKEMURA Haruo

IPSJ SIG Notes. CVIM 164 ( 82 ) 115 - 122 2008.8

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

CiNii Article

CiNii Books

researchmap

Human pose estimation using volumetric features and boosting approach

TAIRA Ryosuke, NAKAZAWA Atsushi, TAKEMURA Haruo

IEICE technical report 108 ( 198 ) 141 - 146 2008.8

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

CiNii Article

CiNii Books

researchmap

Example Based Approach for Human Pose Estimation Using Volume Data and Graph Matching

TANAKA Hidenori, NAKAZAWA Atsushi, TAKEMURA Haruo

The IEICE transactions on information and systems 91 ( 6 ) 1580 - 1591 2008.6

Using the Sequencing of 3D-Video to Create a Movie with an Arbitrary Viewpoint that can be Played Indefinitely

HATTORI YUICHI, NAKAZAWA ATSUSHI, TAKEMURA HARUO

IPSJ SIG Notes. CVIM 2008 ( 3 ) 41 - 48 2008.1

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

3D video with an arbitrary viewpoint, generated by 3D reconstruction of images captured by multiple cameras, is currently actively researched as a new type of media. In our research, we propose a method to connect different 3D video clips in an attempt to realize interactive 3D video. Conversion of time sequence 3D mesh shapes into skeleton data enables us to find frames that can connect each clip, determine a correspondence between the two mesh shapes, and generate an interpolated path of vertices. The target of our method is a multiple joint model such as a human body, so we can create a human's loop, and naturally switch to a random action at any time. So we can apply this to the creation of interactive media, for example, an object changes when the watcher "touches" it in the virtual immersive space.

CiNii Article

CiNii Books

researchmap

Other Link： http://id.nii.ac.jp/1001/00051926/

Using the Sequencing of 3D-Video to Create a Movie with an Arbitrary Viewpoint that can be Played Indefinitely

HATTORI YUICHI, NAKAZAWA ATSUSHI, TAKEMURA HARUO

IEICE technical report 107 ( 427 ) 41 - 48 2008.1

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

CiNii Article

CiNii Books

researchmap

Optimized rendering for a three-dimensional videoconferencing system Reviewed

Rachel Chu, Susumu Date, Seiki Kuwabara, Atsushi Nakazawa, Haruo Takemura, Daniel Tenedorio, Jürgen P. Schulze, Fang-Pang Lin

Proceedings - 4th IEEE International Conference on eScience, eScience 2008 540 - 546 2008

Language：English

Industry widely employs the two-dimensional videoconferencing system as a long distance communication tool, but current limitations such as its tendency to misrepresent eye contact prevent it from becoming more widely adopted. We are exploring the possibility of a three-dimensional videoconferencing system for future interactive streaming of point cloud data, and present the preliminary research results in this paper. We have tested thus far with one sender and one receiver, using pre-recorded data for the sender. The sender, encircled by high-definition cameras, stands and speaks in a room. A cluster of computers reconstructs each frame of the camera images into a 3D point cloud and streams it across a high-speed, low-latency network. On the receiving end, a splat-based Tenderer employs a new algorithm to efficiently resample the points in real-time, maintaining a user-specified frame rate. Parallel hardware projects onto multiple screens while head tracking equipment records the viewer's movements, allowing the receiver to view a stereoscopic 3D representation of the sender from multiple angles. We can combine these visuals with appropriate use of multiple audio channels to forge an unparalleled virtual experience. This next step towards immersive 3D videoconferencing brings us closer to empowering worldwide collaboration between research departments. © 2008 IEEE.

DOI： 10.1109/eScience.2008.42

Scopus

researchmap

Next-generation Teaching and Learning Platform for Higher Educational Institutions(<Special Issue> Application and Aggregation of Learning Objects and Learning Data)

KAJITA Shoji, KAKUSHO Koh, NAKAZAWA Atsushi, TAKEMURA Haruo, MINOH Michihiko, MASE Kenji

Japan journal of educational technology 31 ( 3 ) 297 - 305 2007.12

Language：Japanese Publisher：Japan Society for Educational Technology

This paper describes Course Management Systems (CMSs) from the viewpoint of Open Source and Open Standard in terms of Software Architecture. CMS is a key information infrastructure for supporting teaching and learning in higher educational institutions. The effective use of CMS requires the integration with Student Information Systems (SIS), e-Library and other information infrastructures, along with strategic planning and sustainable development. In this paper, we also introduce our latest activities to develop next-generation CMS since 2004.

DOI： 10.15077/jjet.KJ00004964294

CiNii Article

CiNii Books

researchmap

高等教育機関における次世代教育学習支援プラットフォームの構築に向けて Reviewed

梶田将司, 角所考, 中澤篤志, 竹村治雄, 美濃導彦, 間瀬健二

日本教育工学会論文誌 31 ( 3 ) 36-45 2007.12

Real-Time Space Carving Using Graphics Hardware(<Special Section>Image Recognition and Understanding)

NITSCHKE Christian, NAKAZAWA Atsushi, TAKEMURA Haruo

IEICE transactions on information and systems 90 ( 8 ) 1175 - 1184 2007.8

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

Reconstruction of real-world scenes from a set of multiple images is a topic in computer vision and 3D computer graphics with many interesting applications. Attempts have been made to real-time reconstruction on PC cluster systems. While these provide enough performance, they are expensive and less flexible. Approaches that use a GPU hardware-acceleration on single workstations achieve real-time framerates for novel-view synthesis, but do not provide an explicit volumetric representation. This work shows our efforts in developing a GPU hardware-accelerated framework for providing a photo-consistent reconstruction of a dynamic 3D scene. High performance is achieved by employing a shape from silhouette technique in advance. Since the entire processing is done on a single PC, the framework can be applied in mobile environments, enabling a wide range of further applications. We explain our approach using programmable vertex and fragment processors and compare it to highly optimized CPU implementations. We show that the new approach can outperform the latter by more than one magnitude and give an outlook for interesting future enhancements.

DOI： 10.1093/ietisy/e90-d.8.1175

CiNii Article

researchmap

Automatic Synthesis of Dance Performance Using Motion and Musical Features

SHIRATORI Takaaki, NAKAZAWA Atsushi, IKEUCHI Katsushi

The IEICE transactions on information and systems 90 ( 8 ) 2242 - 2252 2007.8

Refinement of the Shape Reconstructed by Visual Cone Intersection using Fitting the Standard Human Model

HATTORI YUICHI, NAKAZAWA ATSUSHI, TAKEMURA HARUO

IPSJ SIG Notes. CVIM 2007 ( 31 ) 147 - 154 2007.3

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

3D human model reconstruction using visual cone intersection between images captured by multiple cameras has been thoroughly researched, but the accuracy of the shape estimation by previous methods is limited due to the use of silhouette images. Increasing the number of cameras can help improve accuracy, but the problems of unsmooth surfaces appearing in the shape data and the disappearance of the surface normal will still remain. In this research, we try to refine this process by reshaping the reconstructed model. First, we estimate a subject's pose from captured human shape data and create a parts-fitted model using standard human model parts prepared in advance. Afterwards, we reshape the parts-fitted model based on the surface normal vector. By doing this, we can obtain an accurate and refined 3D human model that maintains the subject's pose and characteristics of his body.

CiNii Article

CiNii Books

researchmap

Other Link： http://id.nii.ac.jp/1001/00052069/

Multilinear analysis for task recognition and person identification Reviewed

Manoj Perera, Takaaki Shiratori, Shunsuke Kudoh, Atsushi Nakazawa, Katsushi Ikeuchi

2007 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS, VOLS 1-9 1415 - + 2007

Language：English Publisher：IEEE

This paper introduces a Multi Factor Tensor(MFT) model to recognize motion styles and person identities in dance sequences. We apply a musical information analysis method in segmenting the motion sequence relevant to the key poses and the musical rhythm. We define a task model considering the repeated motion segments, where the motion is decomposed into person invariant factor task and person dependant factor style. We capture the motion data of different people for a few cycles, segment it using the musical analysis approach, normalize the segments using a vectorization method, and realize our MYT model. The experiments are conducted according to two approaches. Various experiments that we conduct to evaluate the potential of the recognition ability of our proposed approaches and the results demonstrate the high accuracy of our model. The recognition results and the motion decomposition with be used in further extending the motion generation process in various styles and for different tasks.

DOI： 10.1109/IROS.2007.4399293

Web of Science

researchmap

ボリュームデータの細線化とグラフマッチングを用いた事例ベース人体姿勢推定

田中秀典, 中澤篤志, 竹村治雄

研究会講演予稿 229 57 - 64 2006.11

Compression of Time Seaquence Volume Data using 3D Hu Invariant Moments

HATTORI Yuichi, NAKAZAWA Atsushi, MACHIDA Takashi, TAKEMURA Haruo

IPSJ SIG Notes. CVIM 2006 ( 51 ) 145 - 150 2006.5

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

Time sequence 3D data has a broad range of application and is a research area expected for the future development. The time sequence 3D data has such a huge data size that data compression is essential when it is transmitted and saved. This paper proposes a compression method by using the time sequence block-matching with extended 3D Hu invariant moment for the purpose of reduction of the data time redundancy. In this method, the data compression ratio is improved by detecting the rotational motion of an object that is not considered in the existing method.

CiNii Article

CiNii Books

researchmap

Other Link： http://id.nii.ac.jp/1001/00052187/

Example Based Approach for Human Pose Estimation using Volume Data and Graph Matching

TANAKA Hidenori, NAKAZAWA Atsushi, MACHIDA Takashi, TAKEMURA Haruo

IPSJ SIG Notes. CVIM 2006 ( 51 ) 131 - 136 2006.5

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

In this paper, we propose a novel marker free motion capture method by using volume data. A volume data is reconstructed from multiple camera views at each frame through visual hull based method. Then, the thinness process is performed to idenfity the structure of the volume. Here, we can get the model skeleton graph in which a body and limbs are expressed as the nodes, and links expresses the connectivity between them. We compare the acquired graph and the graphs in the Model Graph Database (MGDB) and find the most similar one. The MGDB contains the example graphs which express the human body postures. Because the nodes of the MGDB graph are labeled according to the body portions, we can know the portions of input graph (skeleton) from the graph matching result. Finally we fit the skeleton and human body portions' model by using the identification results of the body portions. The experiment result shows the validity our approach.

CiNii Article

CiNii Books

researchmap

Other Link： http://id.nii.ac.jp/1001/00052185/

Report on the 10th IEEE International Conference on Computer Vision (ICCV2005)

FURUKAWA Ryo, KATO Takekazu, KAWASAKI Hiroshi, MITA Takeshi, MIYAZAKI Daisuke, NAKAZAWA Atsushi, SATO Tomokazu, SUGAYA Yasuyuki, UTSUMI Akira, SUGIMOTO Akihiro, SATO Yoichi, WADA Toshikazu

IEICE technical report 105 ( 674 ) 249 - 258 2006.3

　 PREV - NEXT 　

< Nakazawa Atsushi >

MISC - Nakazawa Atsushi