著者
知念 賢一 吉田 豊一 山口 英 香取 啓志
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.51, no.6, pp.925-930, 1997-06-20
被引用文献数
6

Recently, the Internet has become a popular communication medium in which various kinds of information can be provided with world-wide coverage. Interactive information services through the Internet are also expected. The TV community is now trying to develop a new service model of TV broadcasting merged with interactive information services like the WWW over the Internet. In this paper, we report an experiment on a multimedia information service about a 14-day baseball championship as an attempt to merge the Internet with TV broadcasting. We designed a new feed system for the Internet since conventional systems are not suitable for feeding multimedia information to the Internet. This paper compares TV broadcasting and information services on the Internet, presents the design concept of our new feed system, and analyzes it through the experiment.
著者
石川 智也 山澤 一誠 横矢 直和
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.60, no.10, pp.1599-1610, 2006-10-01
被引用文献数
2 1

The advent of high-speed network and high performance PC_s has prompted research into networked telepresence, which allows a user to see a virtualized real scene in remote places. View-dependent representation, which provides a user with arbitrary images using an HMD or an immersive display, is especially effective in creating a rich telepresence. The goal of this study is to create a novel view telepresence that enables a user to control the viewpoint and view-direction by virtualizing real dynamic environments. We describe a novel method of generating views that uses image-based rendering techniques from multiple omni-directional images captured from different positions and that evaluates image quality using a simulated environment. We also describe our prototype system and an experiment with the novel view telepresence that used the system in a real environment. Our prototype novel view telepresence system constructs a virtualized environment from real live videos. The system synthesizes a view based on the user's viewpoint and view-direction as measured by a magnetic sensor attached to an HMD and presents the generated view on the HMD. Our system can generate a user's view in real-time by presenting corresponding points and estimating camera parameters in advance.
著者
鈴木 一隆 豊田 晴義
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.61, no.12, pp.1774-1778, 2007-12-01
被引用文献数
4 3

High-speed, minute eye movements such as ocular microtremors (OMTs), flicks and saccades contain important information about the health of the subject. Since OMTs have a frequency as high as 100Hz and minute angular displacement as low as 0.01 degree, measuring the speed and position of OMTs with adequate accuracy under non-contact conditions in the long-term is difficult. To measure a living sample stably and reliably, we have developed a novel system for measuring eye movement by combining an intelligent vision system and the corneal reflex method. The intelligent vision system has both high-speed and real-time parallel image processing capabilities, which measure to a high level of precision at 1KHz. We have carried out basic experiments using a simulated eye in conjunction with experiments on OMTs and saccade measurements. The results of these experiments confirmed that such a system is sufficiently accurate.
著者
福嶋 慶繁 松本 健太郎 圓道 知博 藤井 俊彰 谷本 正幸
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.4, pp.564-571, 2008-04-01
被引用文献数
3

We have developed an effective rectification method that has a pixel error of less than 0.25 of a pixel for use in two-dimensional multi-camera arrays. Rectification and correction are important for free view-point imaging and multi-view coding in the field of image-based rendering (IBR). In the computer vision field, a number of rectification methods for binocular/trinocular stereo camera setups have been develpopd and used. However, not many multi-camera array rectifications methods have been developed.
著者
西掘 研人 竹内 義則 松本 哲也 工藤 博章 大西 昇
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.7, pp.1086-1097, 2008-07-01
被引用文献数
1 1

We propose a new method for determining the correspondence between audio and visual events based on the selective attention which is observed in a living organism. For auditory attention, our system recognizes a target sound in noisy environments using an auditory filter that focuses on a specific period corresponding to visual events. It was confirmed that the success rate of correspondence between audio and visual events increased from 78.6% without the auditory filter, to 95.2% with the auditory filter. This certifies the importance of visual information for auditory attention. We also realized a visual attention of localizing a sound source and adjusting the line of sight and the iris of a camera when the system perceived a sound, but could not obtain visual information. The success rate of experiments based on visual attention is 93.3%. These results show the effectiveness of the proposed method.
著者
島本 洋 三橋 政次 峯 忠己 船津 良平 橘川 俊哉 三谷 公二 浜田 浩行 羽鳥 好律
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.7, pp.1104-1109, 2008-07-01
被引用文献数
5 1

We have developed a prototype single-chip color HDTV camera for broadcasting. This camera uses a 1.25-inch 8.9-M-pixel CMOS image sensor for obtaining enough resolution characteristics and signal-to-noise ratio (SNR) in a singlechip color system. The camera has an HD-SDI output and a body weight of only 1.4 kg (3.1 lb). The measured resolution was more than 35% on a modulation transfer function (MTF) (45% on a square wave response) at 800 TV-lines. The sensitivity was an f-stop of 11 at 2000 lux with a dynamic range of 400%, and the SNR was 60 dB. These characteristics were as high as those with conventional 2.2-M-pixel 3-CCD HDTV cameras. The measured color difference (delta-E) between this camera and a 3CCD camera on a CIELAB color space was about 9.3. Our simulation showed that the delta-E was about 1.8 with a linear matrix color correction.
著者
海老澤 嘉伸 中島 彩
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.7, pp.1122-1126, 2008-07-01
被引用文献数
4 11

In general, the relative position between the centers of the pupil and corneal reflection are used for precise eye-gaze detection. Recently, for stable pupil detection under various ambient light conditions, the pupil has been detected by the difference between the bright and dark pupil images. However, due to the difference in acquisition timing between the bright and dark images, the precision of the pupil position easily decreases when the head moves. We propose a compensation method which differentiates those pupil images after shifting one of the images so that the positions of corneal reflections in both images should match, and show its effectiveness.
著者
三科 智之 山本 健詞 大井 隆太朗 奥井 誠人
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.7, pp.1132-1137, 2008-07-01
被引用文献数
5

We describe how to reduce interference between elemental images suitable for production of holograms during integral photography pickup that uses a lens array composed of typical convex lenses used as elemental lenses. We propose the use of an optical system that consists of both a field lens and an aperture that is located in the focal plane of the field lens. The system is arranged between the lens array and the camera used for taking elemental images. Diffraction patterns in the focal plane of the field lens of beams going through elemental lenses were calculated with the Fresnel-Kirchhoff diffraction formula. Consequently, we clarified that the system was able to limit the incident angle of the beam passing through the elemental lenses with the size of the aperture located in the focal plane. In addition, we carried out an optical experiment and proved that using the optical system reduced the interference between the elemental images.
著者
大竹 剛 真島 恵吾 合志 清一 藤井 亮介 伊藤 浩 鈴木 光義 高井 重典 寛座 武智
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.8, pp.1292-1300, 2008-08-01
被引用文献数
1

As digital broadcasting has become widespread and the available bandwidth for Internet access has become wider, more high-quality digital content has been distributed to users, making illegal copying a critical issue in the content business. New technology for identifying illegally distributed content is necessary to deter it. We developed two methods for identifying video content: the extraction of characteristic information from video and the use of phase-only correlation. A system was developed to apply these methods to digital TV programs. Experiments on four weeks of broadcast content gave good results, which suggests that a combination of the two methods would enable us to identify video content distributed illegally over the Internet precisely and quickly.
著者
武藤 秀樹 須川 成利
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.8, pp.1319-1325, 2008-08-01
被引用文献数
1 1

A novel wave optical simulation method (a localized boundary element method) has been developed. This method enables us to execute 3-D wave optical simulation with much smaller memory space and much shorter calculation time than conventional boundary element methods or FDTD(Finite Difference Time Domain) methods. The color shading characteristics of inner lens structures in CCD image sensors and the light-gathering power and crosstalk of light waveguide structures in CMOS image sensors can be analyzed by this method. This method was found to be powerful and useful for wave optical analysis of image sensors.
著者
井上 光平 浦浜 喜一
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.8, pp.1326-1329, 2008-08-01
被引用文献数
1 1

Non-photorealistic rendering(NPR) methods are presented for generating halftone images using a centroidal Voronoi diagram(CVD) and its Delaunay triangulation(DT). On the basis of the approximation of a CVD by hexagonal lattice, relations of brightness between an input image and its CVD and DT are derived and applied to brightness correction of halftone images. Experimental results showed that the halftone images generated by the proposed methods reproduced the continuous tone of the input image well.
著者
佐藤 浩司 青木 工太 長橋 宏
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.62, no.8, pp.1330-1332, 2008-08-01
被引用文献数
1

Motion capture data is widely used to easily reproduce natural-looking human motions. However, it is very difficult to search a huge motion database for any particular human motion. A variety of motion retrieval techniques that enable users to find accurately corresponding motions have been proposed. We developed a construction and retrieval technique for the human motion database that considers the overall features of motion by using a dimension reduction technique and a machine learning algorithm. This technique enables efficient retrieval of a set of motions similar to the query. It is also possible to apply our technique to motion synthesis by creating a hierarchical motion graph based on the human motion database.
著者
植松 裕子 斎藤 英雄
出版者
一般社団法人映像情報メディア学会
雑誌
映像情報メディア学会誌 : 映像情報メディア (ISSN:13426907)
巻号頁・発行日
vol.60, no.6, pp.909-919, 2006-06-01

We describe a new on-line AR system, using multiple planar markers placed at arbitrary positions and directions. In most marker-based approaches, the markers' 3D positions and directions need to be measured in advance because they need to define a virtual object's coordinate in one fixed 3D coordinate system. However, such measurement is a very difficult and time-consuming task especially when the markers are in a complex arrangement or are placed in a wide space. Thus, we created a new on-line AR system that does not have any information on the geometrical relationship of the markers. In our registration method, we construct projective 3D space using two reference images, in which we can estimate the geometrical relationship of the multiple markers. By defining the virtual object's coordinate in projective space, this coordinate can be represented in one fixed 3D coordinate system for each frame. We performed some experiments to demonstrate the effectiveness of our system. A virtual object was superimposed onto input image sequences captured with a handy camera. Because the multiple markers were placed in a wide area, the virtual objects could easily move around the real world. Moreover, the registration could be carried out stably because the markers face various directions and because some of them can always be recognized from most viewpoints.