All issues
Predecessor
21.7
Displaying 1-14 of 14 articles from this issue
- |<
- <
- 1
- >
- >|
-
Article type: Cover
Pages Cover1-
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (17K) -
Article type: Index
Pages Toc1-
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (70K) -
Jun Yang, Dili Zhang, Noboru Ohnishi, Noboru SugieArticle type: Article
Pages 1-8
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe discuss the uniqueness of 3-D shape recovery of a polyhedron from a single shaded image. We show that multiple convex (and/or concave) shape solutions usually exist for a polyhedron of which three or more facets meet at an apex (like a pyramid). It has discussed how to check and avoid the superstrictness for a complex polyhedron. However, there is different problem for a simple polyhedron without superstrictness. Horn has shown a numerical example in which two convex and two concave shape solutions exist for a trihedral corner. We describe that multiple shape solutions exist for a pyramid analytically. We also show the shape can be uniquely determined by using interreflections as a constraint to limit the shape solution for a concave polyhedron.View full abstractDownload PDF (807K) -
Dili Zhang, Yoshihiko Nomura, Seizo FujiiArticle type: Article
Pages 9-14
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper addresses the calibration accuracy of a TV camera parameter, i.e., the principal distance. Fiducial marks on a grid plane are translated in the direction vertical to the plane, and the parameter is calibrated from the fiducial-mark image coordinates by using a direct nonlinear minimization. Considering the propagation from the image coordinates measurement errors, we examine the errors of the calibrated parameters and those of the visual angles in 3D space from cameras. Thus, an optimal setup condition with minimum visual-angle error was found among various fiducial mark setups. The errors in the optimal condition are also shown by figures : it enables us to estimate approximate values for any cameras with different principal distances.View full abstractDownload PDF (424K) -
Kenta Hori, Masakazu Imai, Takeshi UnoArticle type: Article
Pages 15-21
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we present the method to detect joint from a pair of pieces that is known to join. Our method doesn't need any special feature of the puzzle pieces. To realize the joint detection method like our method, it is necessary to compare a pair of numerical series that represents the shape of each piece, respectively. In our method, we detect candidates of the joint with clustering pairs of partial contours using MSA (Matching Score Array). MSA is a 2-D array whose components are matching features. The proposed method has been experimented with a pair of potsherds excavated at historical sites.View full abstractDownload PDF (719K) -
JaeMoon Chung, Noboru OhnishiArticle type: Article
Pages 23-30
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSFirst an analysis on geometrical properties of local symmetry of planar shapes is given. Based on the analysis, a new shape descriptor called the chain of circles (CoCs) is defined herein. Then it is shown that the CoCs representation obtained can be controlled in a hierarchical manner which appeals to intuition. The coarse-to-fine hierarchy makes matching of shapes possible with less computational complexity and greater robustness to noise, rotation and scaling of shapes. The dissimilarity vector calculated in the matching may facilitate the searching process in the digital library. The capability of the suggested method is shown by matching several non-formal shapes such as map images.View full abstractDownload PDF (731K) -
Yoshinori Takeuchi, Noboru Ohnishi, Noboru SugieArticle type: Article
Pages 31-38
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we describe an active vision that fixates a saliency in a scene. We define intensity information to determine saliency in the image. The intensity information is based on an information theory, and is theoretically derived. We show that the intensity information is an appropriate measure of image complexity by appling it to various images. We developed an active vision system that selectively attends to the part of image complexity. The developed system has a zooming function emulating the function of the human fovea. The system calculates the intensity information and constructs a saliency map. The system focuses on a point with large saliency, and zooms in. Experimental results demonstrate that the system can efficiently search the saliency in a real scene.View full abstractDownload PDF (1293K) -
Atsushi FUKUDA, Yoshimichi OHTSUKA, Hiroyuki IMAIZUMI, Koichi HAMADA, ...Article type: Article
Pages 39-46
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, foreign FM broadcasting radio waves reflected by the sporadic E-layer have increased interference with Japanese Television Broadcasting. The authors presented how to detect the frequencies of the waves interfering with TV signals using FFT and how to reduce them. By integrating the FFT coeffcients of one scanning line over a field, we can get a smooth spectrum which has some peaks at the interfered frequencies. When supressing the amplitude of the FFT coefficents at the interfered frequencies, we can get TV signals much less interfered. This paper reports on the experimantal equipment for reducing radio interference using the above method. And the efficency is proved by the equipment. Futhermore, we estimated the calculation precision of FFT/IFFT by a computer simulation. As the result of it, a 5 bit accuracy is enough for the cos/sin table assuming equipment loss of 1 [dB].View full abstractDownload PDF (826K) -
Hiroaki Maki, Miki Haseyama, Hideo KitajimaArticle type: Article
Pages 47-52
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper proposes improved cut detection for image sequences using pattern recognition. A method reported before is based on image segmentation and classification of the regions. This method detects a cut by the disappearance of classes. But pattern recognition used for the classification is influenced by interframe motion. To cope with this problem, this paper considers features such as motions and region sizes. Effectiveness of the proposed method is verified by experiments.View full abstractDownload PDF (638K) -
Ryoichi KAWADA, Atsushi KOIKE, Shuichi MATSUMOTOArticle type: Article
Pages 53-59
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSConventional block-based motion picture coding schemes have the problem that motion-compensated prediction efficiency is reduced in a block which contains mixed movements. In order to prevent this, segmentation-based motion picture coding has much potential, and segmentation and motion estimation play important roles. These two processes, therefore, should essentially be performed in close relation with each other, but up to now no full study from this point of view has been made. In this paper, for the purpose of application in motion-compensated predictive coding, the authors propose a scheme for segmentation and motion estimation which integrates the two processes.View full abstractDownload PDF (759K) -
Hisashi MIYAMA, Seiki INOUEArticle type: Article
Pages 61-68
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe have studied the way of tracking human and object motions. In order to track and analyze human and object motions on picture, it is important to divide the picture into individual objects. Moreover, it is necessary to divide two subjects which are close or overlapping each other, also having similar colors and shapes in scenes, and to track the objects which change their shapes. To cope with these difficulties, multiple picture analysis algorithms must be combined so that objects with similar colors and shapes can be distinguished. It is necessary to autonomously renew the information about objects, such as human posture and shape. In this paper, a method is proposed in which autonomous distributed architecture is used, and which combines multiple picture analysis algorithms and integrates analysis algorithm results. The method renews the model autonomously. The relation between the model and a real picture is expressed by probability, and decisions are introduced by the integration of probability. Moreover, we made the decision that model renewal operation is necessary or not. We applied our algorithm to numeral analysis of human body position in the same background domain, a scene of soccer, and report the results herein.View full abstractDownload PDF (1594K) -
Hirotsugu KINOSHITAArticle type: Article
Pages 69-75
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we propose a new encryption method which have characteristics that are much more compatible with the current MPEG coding technology. The encryption methods for the macro block layer and the slice layer are combined to incrcase the conceal-ness. Furthermore, The keys for the encryptions are generated from the secret information which is shared with the sender and the receiver by the conventional cryptography.View full abstractDownload PDF (837K) -
Sei NAITO, Ryoichi KAWADA, Shuichi MATSUMOTOArticle type: Article
Pages 77-82
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRegarding motion compensation for moving pictures, various methods have been proposed so far. In these proposals, merits and demerits regarding performance are discussed for each method, but they are not necessarily discussed under fare conditions. In this paper, we investigate these 4 methods as representative motion compensation methods, "block matching", "motion compensation using affine transformation, ""warping prediction"and"global motion compensation". Then we evaluated the performance of those methods, applying the typical coding method, MPEG-2,for the input of common test sequences. As a result, we show that motion compensation methods which have been proposed to improve the performance are not superior to the conventional block matching when they are used in conjunction with MPEG-2.View full abstractDownload PDF (568K) -
Article type: Appendix
Pages App1-
Published: February 04, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (72K)
- |<
- <
- 1
- >
- >|