Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：ACM  

This study proposes an extended geographical database based on photo shooting history to enable the suggestion of candidate captions to newly shot photos. The extended geographical database consists of not only subject positions but also the likely shooting positions and directions estimated using the histories of the shooting positions and directions of the subjects. A user can add a caption to a photo by selecting an appropriate one from the candidate captions. The candidate captions are acquired using the shooting position and direction as a key to the extended geographical database. In this paper, we present the results of experiments for constructing the extended geographical database using a prototype system. Copyright 2008 ACM.

DOI： 10.1145/1386352.1386381

Scopus

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/civr/civr2008.html#conf/civr/IwasakiKYY08

Language：Japanese   Publisher：電子情報通信学会  

CiNii Books

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：TSI PRESS  

Recently a number of catadioptric video-rate oninidirectional imaging sensors with 360-degree of horizontal view have been developed. Among those onmidirectional video cameras, ones which satisfy the single-viewpoint constraint are suitable for emerging interactive media-oriented applications, because perspective re-projection onto any display or screen sufaces can be computed from captured omnidirectional images. This paper describes some applications of onmidirectional video camera HyperOmni Vision to visual information media, especially focussing on our recent activities including telepresence, telerobotics, and video surveillance.

DOI： 10.1109/WAC.2002.1049514

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper presents a new method for constructing a virtual environment from a real world scene, in which users can look around the scene with binocular stereoscopic sensation. The approach is based on generating novel views at arbitrary positions from a sequence of omnidirectional images in real time. A sequence of omnidirectional images is captured by a moving omnidirectional video camera with a known velocity in a static real environment. In the proposed method, a fixed length of the image sequence is temporarily buffered to generate arbitrarily directional binocular stereo images combining captured rays. Telepresence systems have been developed for showing the feasibility of the method, assuming the omnidirectional camera running along a straight line with a constant speed. Experiments have shown that a user can look around a remote scene in the distance from moving viewpoints with rich 3D sensation without significant time delay.

Web of Science

researchmap

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

CiNii Books

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE COMPUTER SOC  

Recently virtual reality (VR) systems have been incorporating rich information available in the real world into VR environments in order to improve their reality. This stream has created the field of mixed reality which seamlessly integrates real and virtual worlds. This paper describes a novel approach to the construction of a mixed environment. The approach is based on capturing the dynamic real world by using a video-rate omnidirectional stereo image sensor. The mixed environment is constructed of two different types of models: (1) texture-mapped cylindrical 3-D model of dynamic real scenes and (2) 3-D computer graphics (CG) model. The cylindrical 3-D model is generated from full panoramic stereo images obtained by the omnidirectional sensor that uses a pair of hexagonal pyramidal mirrors and twelve CCD cameras. A prototype system has been developed to confirm the feasibility of the proposed method, in which panoramic binocular stereo images of the mixed environment are projected on a cylindrical immersive display depending on user's viewpoint in real time.

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE COMPUTER SOC  

The recent progress in computer graphics has made it possible to construct various virtual environments such as urban or natural scenes. This paper proposes a hybrid method to construct a realistic virtual environment containing an existing real scene. The proposed method combines two different types of 3-D models. 3-D geometric model is used to represent virtual objects in user's vicinity, enabling a user to handle virtual objects. Texture mapped cylindrical 2.5-D model of a real scene is used to render the back-ground of the environment, maintaining real-time rendering and increasing realistic sensation. Cylindrical 2.5-D model is generated from cylindrical stereo images captured by an omnidirectional stereo imaging sensor. A prototype mixed reality system has been developed to confirm the feasibility of the method, in which panoramic binocular stereo images are projected on a cylindrical immersive projective display depending on user's viewpoint in real time.

Web of Science

researchmap

Language：Japanese   Publisher：Institute of Systems, Control and Information Engineers  

DOI： 10.11509/isciesci.43.2_110

CiNii Books

CiNii Research

researchmap

Language：Japanese   Publisher：社団法人映像情報メディア学会  

CiNii Books

CiNii Research

researchmap

Language：Japanese   Publisher：社団法人映像情報メディア学会  

CiNii Books

CiNii Research

researchmap

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

To recognize the whole three-dimensional environment, it is necessary to take images of real scenes in all directions and to obtain depth information from the two-dimensional images. In this report, we present an omnidirectional stereo method. Our method can detect disparities between two omnidirectional images. We first take two omnidirectional images at different heights using the omnidirectional sensor HyperOmni Vision, and then detect corresponding points between a pair of omnidirectional stereo images and obtain three-dimensional information. Experimental results show the feasibility of the proposed method with real omnidirectional images.

DOI： 10.11485/itetr.21.6.0_99

CiNii Books

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer Verlag  

In this paper, an omnidirectional stereo method is presented. The method first takes a pair of omnidirectional images at different heights using the omnidirectional imaging sensor, HyperOmni Vision, and then detects corresponding points between the pair of images and finally obtains three-dimensional information of a 360-degree view around the sensor.

DOI： 10.1007/3-540-63930-6_133

Scopus

researchmap

Event date： 2004.8

Language：Japanese  

researchmap

Event date： 2004.3

Language：Japanese  

researchmap

Event date： 2002.9

Language：Japanese  

researchmap

Event date： 2002.9

Language：Japanese  

researchmap

Event date： 2002.9

Language：Japanese  

researchmap

Event date： 2001.8

Language：Japanese  

researchmap

Event date： 2000.9

Language：Japanese  

researchmap

Event date： 2000.9

Language：Japanese  

researchmap

Event date： 2000.9

Language：Japanese  

researchmap

Event date： 2000.3

Language：Japanese  

Recently, there is an increasing demand for virtualizing an outdoor real environment. However, such a task usually requires lots of manual operations to determine correspondences among multiple images. In this paper, a method for automatic reconstruction of an outdoor environment is proposed. The method uses a GPS(global positioning system)receiver and gyro sensor to measure an exact location and orientation of a camera for image acquisition. By using these location and orientation data, stereo images, of an outdoor environment are rectified for stereo measurement. Then the method merges multiple partial models of an outdoor environment to construct an entire 3-D model. In this process, the proximity constraint is imposed on adjoining faces to compensate for errors caused by the multiple stereo measurements.

researchmap

Event date： 1999.9

Language：Japanese  

researchmap

Event date： 1999.9

Language：Japanese  

researchmap

Event date： 1997.9

Language：Japanese  

researchmap

Event date： 1997.3

Language：Japanese  

researchmap

Event date： 1996.9

Language：Japanese  

researchmap