Reports of the Technical Conference of the Institute of Image Electronics Engineers of Japan
Reports of the 224th Technical Conference of the Institute of Image Electronics Engineers of Japan
Showing 1-23 articles out of 23 articles from the selected issue
Date: March 17, 2006 Location: Nagasaki University
9:00-10:20 Chair :Minoru ITO, Kogakuin University
  • Soutarou TABATA, Kei IWASAKI, Saeko TAKAGI, Fujiichi YOSHIMOTO
    Session ID: 05-07-01
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Although there are many information retrieval systems, few systems for retrieving in information of objects in real worlds can be used in mobile environments. In this paper, we describe an image retrieval system of flowers that can be used easily in mobile environments. Using a mobile phone, we transmit a flower image, simple characteristics of the flower and its positional data to an image retrieval server, and we confirm the result of retrieval using the mobile phone. We improved the search system by changing the database and an extraction range of a color, and by introducing narrowing down by size. The best result was provided when the search system used a database of plural flowers, did narrowing down by size, and extracted a characteristic of the color from the whole flower domain. The result of our experiments shows that the percentage getting the target image in the top ten was 95.1%.
    Download PDF (787K)
  • Qiang DONG, Kei IWASAKI, Saeko TAKAGI, Fujiichi YOSHIMOTO
    Session ID: 05-07-02
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we propose an image retrieval system of fishes using mobile equipment that can be operated easily, anywhere. In this system, we take a picture of a fish and transmit the fish image to a server machine with a cellular phone. A database of fishes in the server machine is searched with the sent image. The features of each fish are registered into the database in advance. The target image is searched by comparing features calculated from the sent image with the ones registered in the database. The result of the retrieval is checked on the cellular phone. We developed a prototype of the proposed system and experimented with images of actual caught fishes. The result of our experiments shows that the percentage of getting the target image in the top ten was about 90%. The target fish data was obtained within about 100 seconds.
    Download PDF (3938K)
  • Yasuhiro HENMI, Sunao KIKUCHI, Shigeru AKAMATSU
    Session ID: 05-07-03
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The Galatea Toolkit, provided free by the Galatea Project, is a software environment for developing an anthropomorphic talking agent which displays facial expressions as well as facial speeches. The toolkit includes a function of constructing a 3D face model of an arbitrary person when only one photograph of the full face of the person is given. The 3D face model, however, reveals a kind of unnaturalness in shape when a sidelong glance is given to it. This paper therefore introduces an attempt to adjust the face model to the 3D shape of an arbitrary person that is densely measured by a 3D range finder, to obtain a talking agent with more natural 3D looking of the person.
    Download PDF (230K)
  • Takahiro MIURA, Kimitsugu KON, Shigeru AKAMATSU
    Session ID: 05-07-04
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes automatic person tracking system using Pan/Tilt/Zoom camera. Generally, frame difference is effective to person tracking but the system using active camera cannot use frame difference, because moving object can not be extracted out of the changing background when camera moves. In this paper, a method of judging whether the camera is moving or not is proposed to calculate the area of motion regions. The system is designed not to use the information of frame difference when camera moves. As a result, about a real time system has been achieved that could track people rapidly moving back and forth.
    Download PDF (217K)
10:30-11:50 Chair :Tetsuro KUGE, NHK Science & Technical Research Laboratories
12:40-14:00 Chair :Mitsuji MATSUMOTO, Waseda University
  • Sumio YOKOMITSU, Hiromichi SOTODATE, Masaaki SATO
    Session ID: 05-07-09
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Surveillance system is now required to cover large and wide area. It is infeasible for human to observe the events in each camera and to search the specific event immediately. In this paper, we proposed a method to calculate human height estimation technology in a single camera so that we can search the specific person from the big image databases. In order to clarify the efficiency of our proposed method, it execute the experiment about the estimation of the human height in the outdoor scene.
    Download PDF (311K)
  • Akira NAGAMORI, Takamune SATO, Izumi SANTO, Akio SUGANO, Minoru ITO
    Session ID: 05-07-10
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes a new approach for the detecting pattern edge positions and roughness of LSI ultra-fine patterns. As LSI pattern lines become narrower, exact estimation of line edge roughness and line width roughness as well as edge positions, becomes indispensable. The proposed approach comprises the function approximation of cross-sectional signals at each edge in a SEM image and the automatic determination of the optimum edge detection region used in the function approximation. The approach enables us to detect edge positions and roughness with high accuracy and high reproducibility.
    Download PDF (142K)
  • Ken INOUE, Ken'ichi MOROOKA, Hiroshi NAGAHASHI
    Session ID: 05-07-11
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently, we frequently see the animation of 3D geometric models in our dailylife. To make these models, they often use some systems based on Inverse Kinematics or Motion Capture. However, it's difficult to make animations of the non-rigid body model with these systems. And so, in this research, a motion generation system is proposed, in which users can make non-rigid body model arbitrarily in the field. Bèzier model's control-points can move independently, and the motion of a model is defined as the assortment of controlpoints's movement. This motion is evaluated by the virtual field which is constructed by some parameters. And we propose the algorithm which a model captures some motion in virtual field by optimizing the assortment of movement of control-points with this evaluation.
    Download PDF (534K)
  • Osamu SHIKU, Keiichi TAMAI, Hirofumi FUKUMOTO, Hideo KURODA
    Session ID: 05-07-12
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a method of character string extraction from scene images. In this method, we extracted a character string searching around a character candidate. The character string extraction process is executed for each character candidate extracted by an existent character extraction process. As a result, character strings corresponding to character candidates are extracted. Then, if some character strings are overlapped, redundant character strings are removed. The character string extraction process is executed using segments therefore, the processing cost is small. We designed a system composed of an existent character extraction process and this proposed character string extraction process. The method was applied to 200 camera-captured scene images involving 723 strings. As a result, 90.7% of 723 strings were extracted.
    Download PDF (804K)
14:10-15:30 Chair : Takanori SENOH, The University of Tokyo
  • - Adaptive Control of Optimal Quantizing Scheme -
    Takahiro FUJITA, Sekie OKA, Akira KUBOTA, Yoshinori HATORI
    Session ID: 05-07-13
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Predictive coding based on interpolative prediction using a correlation between all the neighboring pixels is expected to realize high prediction performance compared with extrapolative prediction based coding. In such an interpolative prediction based framework, optimal quantization schemes have been investigated in such a way that difference signals are quantized to minimize quantization errors through their cancellation. In this paper, we present a novel adaptive control method by re-ordering the quantization in this optimal quantization and evaluate the effect on the coding error power. We also discuss extension of the presented adaptive control method to 2-dimensional interpolative prediction.
    Download PDF (381K)
  • Toshihiro AKIYAMA, Motoyoshi FUMIAKI, Osamu UCHIDA, Shohachiro NAKANIS ...
    Session ID: 05-07-14
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we propose a digital watermarking method for color Images based on human visual system using wavelet transform. We embed a watermark into both the luminance component and the chrominance component by using two different algorithms.
    Download PDF (576K)
  • Yukiko NAKAMURA, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 05-07-15
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a system for displaying a TV program with an icon which represents corresponding genres of each program. This system assigns the most suitable icon to each program by using text data contained in the electronic program guide (EPG). We supposed that the process for assigning a suitable icon to a TV program can be replaced to a process for classifying text data (title and description of the program) for explaining the program to a suitable genre. Then we employ the naive Bayes classifier for the classification, because we can prepare training data for the classifier with enough size through fairly easy way. This comes from following two reasons: the EPG is distributed regularly and it is suitable for computation processing and storage of the processing result. We developed a prototype in which we implement the proposed method and applied it to the actual TV program.
    Download PDF (471K)
  • Yuriko TAKASE, Shinji IKARUGA, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 05-07-16
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The community planning, such as the redevelopment of a central urban area, has become popular. Stake holder of the redevelopment, government officers, and people from academic background participate together in the workshop for discussing the community planning. In this paper we propose two kinds of systems by which we can build a virtual 3D space and freely move there, in order to aid smooth decision making at the workshop. Both of them utilize the merit of the VRML (Virtual Reality Modeling Language). In concretely, we developed a system which is characterized on functions for comparative study for using on the examination phase of a landscape plan. We also developed the other system which is characterized on functions for giving a virtual experience of future shape of the region which will appear after the completion of redevelopment for using on the final phase. We confirmed the usefulness of both systems through using them in actual workshops.
    Download PDF (572K)
15:45-17:05 Chair : Yoshinori HATORI, Tokyo Institute of Technology
  • Shiro OZAWA, Takao ABE, Noriyuki NARUTO, Itaru KAMIYA, Atsunobu NARITA ...
    Session ID: 05-07-17
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    3D image has been interested in various fields. Moreover, the image contents came to be communicated actively on the network. However, there is little chance that 3D image is exchanged on the network. We think the cause to be in the place where the ordinary user cannot readily make contents because photography of 3D image is difficult and it needs expensive hardware. Then we have developed the system for taking and distributing 3D images in real time. In this paper, we confirmed that the utility value of live 3D image was verified by the system's that made it for trial purposes examining the application, and there was a possibility that the range of use of 3D image expands.
    Download PDF (500K)
  • - as a candidate for glass-free 3D image generation -
    Takanori SENOH, Terumasa AOKI, Hiroshi YASUDA, Takuyo KOGURE
    Session ID: 05-07-18
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Three-dimensionalization of digital images has been investigated seeking for further reality improvement. A most feasible and stable technology based on the binocular disparity and convergence requires stereo-glasses. However, there is still a strong desire for glass-free method in order to ease the stress to viewers caused by glasses or imperfect 3D sensation. In this report, Space-Sampling method is investigated for glass-free 3D images. With layered image sensor device, Space-Sampling method captures 3D image projected through a lens. Layered display panels reproduce real 3D image with depth. Transparent image sensors, layered display panels, depth resolution analysis and blur elimination problems are studied.
    Download PDF (963K)
  • Jun LIU, Mitsuji MATSUMOTO
    Session ID: 05-07-19
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    SIP protocol has been the focus of much recent effort by IETF standards bodies. While media quality associated with SIP has been studied a lot, the amount of time of setting up the multimedia session in IP-based network received less attention. This paper outlines a simulation study of SIP session setup delay. The focus is signaling setup delay and setup throughput differences by using UDP, TCP and SCTP as transport method for SIP. Results show their appropriateness in the conditions of packet loss and heavy load.
    Download PDF (604K)
  • Tatsuya TSUJI, Takahiro KITAUCHI, Saeko TAKAGI, Noriyuki MATSUDA, Masa ...
    Session ID: 05-07-20
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Currently, there is no system that can improve users' skill to draw pictures, although many digital tools for pictures are provided. We have been paying attention to pencil sketch which is a basis of art activities. We developed a study support system for beginners' pencil sketches of a plate and a glass. The motif, a plate and a glass, is good for learning basic sketches. It is, however, not interesting for learners usually. We propose a system for beginners' pencil sketches of fruit. The system evaluates sketches in each phase of outline and shading the same as the previous system. In this paper, we explain the details of the system for outline sketches of fruit which has different processing from the previous system. We also report the result of the evaluation experiment.
    Download PDF (2528K)
17:15-18:15 Chair :Hidekazu SEKIZAWA, TOSHIBA TEC Document Processing Systems Co.,Ltd.
  • - Bases for the Corresponding Color Reproduction -
    Eiichi NIINOMI, Johji TAJIMA
    Session ID: 05-07-21
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes object color changes under various daylights. The changes were estimated by numerical experiments using Standard Object Colour Spectra Database (SOCS). First we found that, when we estimate the color under the second illumination with a correlated color temperature T' from the color under the first illumination with another correlated color temperature T using object spectral reflectance estimation, the estimated color is very stable independent of metamers. This fact allows us to define the incomplete chromatic adaptation in terms of the correlated color temperature of daylights. Second, we found that the incomplete chromatic adaptation can be symmetrically expressed using a quantity M whose unit is mired (micro reciprocal degree). These results could be the bases for a theory of corresponding color reproduction under different daylights.
    Download PDF (291K)
  • Nobuhito MATSUSHIRO, Noboru OHTA
    Session ID: 05-07-22
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    More than one hundred years ago, von Kries model has been developed for a description of the chromatic adaptation phenomena of the human visual system, and various chromatic adaptation models based on von Kries model have been developing in today. However, there remain some important issues related to von Kries model which have not been clarified up until now. Succeeding to the first report, the theorems derived for subtractive color mixture problems by us will be applied to the important issues related to von Kries model.
    Download PDF (191K)
  • Tetsuro KUGE
    Session ID: 05-07-23
    Published: March 17, 2006
    Released: November 07, 2006
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
feedback
Top