Reports of the Technical Conference of the Institute of Image Electronics Engineers of Japan
Reports of the 200th Technical Conference of the Institute of Image Electronics Engineers of Japan
Showing 1-38 articles out of 38 articles from the selected issue
  • Mitsuhiro SATO, Naohide YAMADA, Madoka HASEGAWA, Shigeo KATO
    Session ID: 02-07-01
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The wavelet transform are known as an effective lossless cording model for the image data. A single filter is usually applied for the image in the wavelet coding method. However, higher compression ratio is not expected for compound images, which consist of both bi-level and natural image, by using only a single filter. We have been studied an adaptive wavelet transform coding by segmenting an image into blocks and applying adaptively proper filters to the divided blocks. Consequently, we have confirmed that it can reduce average code length compared with the conventional wavelet transform coding. However, since the filter is chosen by single lifting ladder circuits, we faced the problem that it has very high computational complexity. In this paper, we propose an adaptive wavelet transform coding that lowers computational complexity by synthetic lifting ladder circuits.
    Download PDF (561K)
  • Yosuke MIYAKE, Madoka HASEGAWA, Shigeo KATO
    Session ID: 02-07-02
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    There are many code words of the same length in Golomb-Rice code. Supposing that the code lengths of the code word CA equals that of the code word CB and they are assigned to prediction error A and B, respectively. Without degeneracy of the prediction errors, the prediction error A may appear and the prediction error B never appear in any prediction value. In such case, even if each of code words CA and CB is used for the prediction error A in the transmitting end, the receiving end can be decode as the prediction error A. If the encoder determines the code word according to the ascriptive information bit, the receiver can find the ascriptive information bit and decode simultaneously the prediction error value. So the code words CA and CB are the same code word length, the multiplex transmission can be realized in the same code length as the conventional predictive coding method.
    Download PDF (421K)
  • Nobutaka SUZUKI, Madoka HASEGAWA, Shigeo KATO
    Session ID: 02-07-03
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Studies of digital museum are making progress in the various fields. Especially, in case of preserving 3D objects, it is necessary to take digital photographs at various angles. Thus, the amount of data becomes huge and a data compression technique should be introduced for archiving them. Moreover, in order to represent art works faithfully, a lossless coding scheme is indispensable. This paper addresses a bidirectional prediction method using motion compensation and rotation compensation for the image sequence of a cylindrical object taken by rotating it on a turntable. As a result of the experiment, the proposed method can reduce the entropy compared with using only motion compensation.
    Download PDF (312K)
  • Akitaka Uchida, Kiyoshi Tanaka
    Session ID: 02-07-04
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    It is very important to control the bit-rate for video communication under a limited capacity of communication channel like wireless communication system because of large amount of data. From this point of view, an adaptive pre-filter to improve the bit-rate in video compression has been proposed, which preserves visually significant portion of the original image and smoothes only insignificant portion actively. In this paper, we propose three kinds of temporal extension of this scheme and verify their perfomance on compression ratio and visual image quality.
    Download PDF (619K)
  • Emi Myodo, Kiyoshi Tanaka
    Session ID: 02-07-05
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we study on a shared watermark generation scheme extended from an image halftoning scheme using a genetic algorithm. In addition to the conventional evaluation functions on the gray-level precision and appropreate contrast near edges, we design and introduce a evaluation function on the watermark visibility to the scheme and generate multiple halftone images embedded (shared) a watermark image. Simulation results show that we can embed a watermark into images with proper parameters keeping high image quality, but clearly decode the embedded watermark by overlapping those images.
    Download PDF (488K)
  • Tomonori TABUSA, Joo kooi TAN, Seiji ISHIKAWA, Kazuhide AKAO
    Session ID: 02-07-06
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes a three-dimensional motion recovery technique in a large field employing multiple-views. Modeling 3-D human motions is requested in various fields such as virtual reality, sport analysis, and medical science. For 3-D motion recovery, direct liner transformation or the factorization based on an extended measurement matrix is generally employed. These techniques normally employ fixed or panning cameras for taking video images. They are therefore difficult in recovering motions in a large field. In this paper, we propose a technique for achieving 3-D recovery of human motions in a large field employing a mobile set of cameras. The 3-D locations of feature points on an object are calculated by factorization and DLT method. Using these two techniques, a cube and a human wide range motion are recovered in a 3-D way. Recovery results are shown, and the accuracy of the recovery is compared. Significance of the technique is also discussed.
    Download PDF (493K)
  • Yoshinori YAMAGUCHI, Hiroyuki MIYAKE, Osamu NISHIKAWA, Tetsuo IYODA
    Session ID: 02-07-07
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes shape measurement of glossy objects by a new range finder, which consists of a co-principal point camera to a projector and a polarization optical system, projects multiple-value structured pattern. The specular light reflected on glossy surface is decreased by a polarization optical system, which consists of a linear polarization projector and polarization filters. Experiments show a high accuracy shape measurement of glossy objects by one shot.
    Download PDF (561K)
  • Yoshikazu TANNO, Norifumi SAGAE, Osamu NAKAGAWA, Fumio HASEGAWA
    Session ID: 02-07-08
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper proposes a visual-based retrieval mechanism that enables users to find which video they want to view from a large-scale archive efficiently. The majority of current research in this field focuses on text-based information analyzing and retrieving. This, however, may become inefficient when the number of video databases is very large. The mechanism proposed in this paper have a multiple video digest display system with a user-friendly interface and a scene-detection-based index that is automatically constructed while performing the content registration process. This allows users to efficiently access the detailed video information. As compared to that of text-based searching, the results show that our proposed mechanism has better performance in terms of effectiveness of target content narrowing.
    Download PDF (481K)
  • Thitiporn LERTRUSDACHAKUL, Akinori TAGUCHI, Terumasa AOKI, Hiroshi YA ...
    Session ID: 02-07-09
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Although the existing videoconferencing systems have been progressed into the level of practical use, they still do not totally fulfill the need for a decision making process. Since the eyes of user cannot always be in contact that will introduce a difficulty in reading an expression between participants in a conference. In this paper, we propose a multiparty videoconferencing system with eye contact based on 2 cameras for IshinDenshinSystem; a videoconferencing system in which participant's image and shared workspace are naturally merged.
    Download PDF (344K)
  • Seiya Miyazaki, Jinhong Shen, Takafumi Yuki, Mitsuru Kaneko, Terumasa ...
    Session ID: 02-07-10
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Many attempts have been done to reduce the burden and the difficulty of the CG production. One of the biggest evolutions is the approach of automatic cinematography, driven by commands or language, instead of traditional direct manipulation. In this paper we present directorial planning engine based on cinematographic rules. This algorithm is being implemented in the DMP (Digital Movie Producer) system that aims at visualizing scenario written in natural language format.
    Download PDF (584K)
  • Shen Jinhong, Seiya Miyazaki, Terumasa Aoki, Hiroshi Yasuda
    Session ID: 02-07-11
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper present a system designed to automate the production of digital movies. In place of a human one, the virtual director automates basic tasks of orchestrating the action in form of the action & dialog and all other ingredients that contribute to the final motion picture like figures and props. It also automates control decisions on aspects of the making of the film such as camera position & movement, sounding, and lighting. The virtual director use cinematic knowledge to achieve user's intentions through knowledge-based approach. We discuss the design of the virtual director and present an example result of its use in the digital movie producer system DMP.
    Download PDF (308K)
  • Mei KODAMA, Masashi MURASAKI
    Session ID: 02-07-12
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We had already proposed "Multimedia Scalability Package:MSP" as structure data in integrated multimedia services. In addition, we studied video delivery systems using MSP data. In this system, updatable scalability and updating processing are used, and MSP function operates these data, play, edit, transmit and so on. We propose a video transcoding method for information updating over network to select high resolution or low resolution video. Video delivery systems and information management methods are also proposed with these transcoding and updating methods. Moreover, video contents delivery systems are indicated using MSP contents. We consider contents tranmition and transformation methods based on MSP data. Especially, video processing methods are explained and these procedures are shown. By simulation results, it is efficient for users to play multi resolution video contents.
    Download PDF (446K)
  • Hiromitsu KIKUCHI, Jiro KATTO
    Session ID: 02-07-13
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    As one of model-based methods for face recognition, a technique called Embedded HMM (Hidden Markov Model) was proposed, in which two or more HMMs are arranged spatially and connected in the fixed direction. This Embedded HMM can provide robust learning and recognition against changes of direction and positions of a face compared with the conventional methods based on PCA (Principle Component Analysis). However, when a face is rotated within an image plane, there is a possibility that face structure could not be expressed well by the Embedded HMM. Therefore, in this paper, we propose a novel Embedded HMM that has two kinds of structures with different direction. Furthermore, in order to improve its performance against partial hiding of a face, such as, existence of glasses, we propose a new recognition algorithm using plausible parts of face image.
    Download PDF (504K)
  • Azusa KAWAMITSU, Tsuyoshi YAMASHIRO
    Session ID: 02-07-14
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The private attestation is required in the various places of the societies and is always asked for high reliability and high convenience. However, in the main attestation method, the demerit such as disclosure, oblivion, forgery, and missing come out. Then, the private attestation using biometrics attracts attention in recent years as a new system. There are some which have already been put practical use in attestation by biometrics, but the picture which serves as the feature with scanning will be received a mental streess. Then, we aimed at realization of the simple private attestation which is possible only by showing up the palm without receiving these stresses in this reserch.
    Download PDF (239K)
  • Tsuyoshi KATO, Hisashi SUZUKI
    Session ID: 02-07-15
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Popular gaze tracking systems and pupillometers expose eyes to near-infrared rays and detect clear images through infrared cameras based on the property that the iris reflects the near-infrared rays while the pupil does not since structure of eyeballs. Some problems are that: near-infrared rays cause users feeling thirsty; long time exposure of infrared rays may cause users cataract; and, infrared rays devices are complicated, etc. Although the best way is to split iris and pupil under visible rays, it is not easy since almost all Japanese have brown irises and black pupils while the surface of eyes reflects the surroundings sensitively. This article reports that examining a possibility of splitting iris and pupil by combining the generalized Hough transform and Fukui's edge extraction method.
    Download PDF (284K)
  • Yasuhiko Watanabe, Toshiyuki Kurahashi, Yoshikatsu Kishida, Yukio Toku ...
    Session ID: 02-07-16
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper evaluates a method for building a meta-database for accessing distributed digital archives. This meta-database provides an integrated accessibility for multiple digital assets in network. A trial system is used to evaluate the method for searching multiple target images from image archives. Experimental results show the improvement of retrieval of images from distributed digital archives through the method.
    Download PDF (287K)
  • Automated Archive of Two-Way Lectures in MPEG2 format
    shunya SUZUKI, Norikazu IWATA, Mei Kodama
    Session ID: 02-07-17
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently, the requirement and importance of contents for distant learning is dramatically increased because of the collaboration among universities and local communities, the variegation of students, distributed campuses. But the cost to produce specially designed e-learning contents for distant leaning is still too expensive for the staffs in universities. To reduce the cost of production, the idea making e-learning contents from archived lectures has been studied since the end of 90s. However, the archived lectures are usually difficult to use as resource of e-learning contents, because common lectures reflect various un-recordable factors like face-to-face communication. We have constructed automated lecture archive system that captures MPEG2 stream of two-way TV lectures from ATM onto huge disk strage and DVD collection, to provide resource for distant learning contents.
    Download PDF (508K)
  • Satoshi Higa, Tsuyoshi Yamashiro
    Session ID: 02-07-18
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Various information overflows and many of the information are shared as a part of the society. To get useful information efficiently is an inevitable condition for as to live. In this study, we paid attention to the black pixels of the letter that is written in the image, and extracted the word that is required. We used the map images as the sample here.
    Download PDF (215K)
  • Application to Description and Synthesis of Moving Images
    Masanobu HAGA, Mitsuhiko MEGURO, Masahide KANEKO
    Session ID: 02-07-19
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently, the rapid progress in computer and network technologies has made the use of moving images more popular. However, it is not a easy task for a user to handle the raw moving images directly; for example, generating moving images by oneself or retrieving arbitrary moving images from large image database. This paper introduces the idea of "tube" which connects the sequence of object shape along the time axis by the third order spline curve. An interactive system to generate animated images based on the idea of tube is proposed. Furthermore a method to generate moving images using texture obtained from real images and a method to describe an object in real moving images by combination of tubes are discussed. Experimental results show the usefulness of the proposed methods.
    Download PDF (502K)
  • Preliminary Experiment on Visual Impression
    Yuichi KOBAYASHI, Masami SUZUKI, Takahiro NAKAI, Norihiro HAGITA
    Session ID: 02-07-20
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In order to realize a co-experience-based communication media, we have studied novel expression which interacts between multi-media. As part of those approaches, we focused on the Haiku's expressive power and started to investigate some expression with multi-media and Haiku. In this paper, we report the analyses the relationship between human subjective impression and color property for "Haiga"-CG and Haiku.
    Download PDF (472K)
  • Kazuya EZURA, Yuji SAKAMOTO, Yoshinao AOKI
    Session ID: 02-07-21
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this report, we propose a method for computer-generated holograms to calculate phenomena that shadows are on objects, an object is illuminated by lights reflected on another objects, and an object penetrates lights. In order to reconstruct a realistic image from computer-generated hologram, to consider these phenomena is indispensable. To calculate them, this method uses shadows, reflection, and penetration of lights on patches of objects. Additionally, we synthesized computer-generated holograms using proposed method, and confirmed the effectiveness of it.
    Download PDF (420K)
  • Tsuyoshi YAMASHIRO, Zensei IHA
    Session ID: 02-07-22
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The Karate and Ryukyu entertainments are the traditional culture that attention to Okinawan prides in the world. The sanshin (three stringed musical instruments) was introduced to Ryukyu from China more than 600 years before (1392 around) and have played the important role of the Ryukyu entertainments. How we can contribute such traditional entertainments? If we could contribute for such traditional entertainments with science-technology, they will be stimulated both of culture and the science-technology. Recently, the amusement robots, witch give us peace of mind, are in interest instead of the production machine. In this paper, I will describe mainly related to the sanshin, such as, musical score, tone of some musical instruments, history, and the mechanisms of the sanshin performance robot "agasaa", which we developd, in the second half of the lecture.
    Download PDF (556K)
  • Make Broadband Business
    Hiroshi YASUDA
    Session ID: 02-07-23
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Speeding up of mobile internet bit-rate, rapid deploying of fiber accesses, and digitalization of terrestrial TV Broadcasting, by those, broadband-ubiquitous communication environment of Japan is coming mature quickly. With this and advanced network function make everyone a broadcaster. Moreover, metadata is now playing a key role in every activities, thus now metadata themselves are now becoming contents. Environment above mentioned forces everyone to make contents, thus we foresee that contents creation and distribution would be a next carrier for industries in 21st century. In this paper, the author would like to describe changes in the background age, and proposes business models to make contents creation and distribution successful businesses.
    Download PDF (236K)
  • Koichiro HAYASHI
    Session ID: 02-07-24
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The copyright system began with the modern printing system, and survived the consecutive technological innovations such as cinematograph, television, copying machine and computer. Actually its scope was expanded every time new inventions were introduced and it is still alive over several centuries. In the 1990's, however, the astonishing growth of the Internet is shaking the foundations of the system. There are usually two contrastive methods to cope with the crisis; one is to strengthen its foundations by amending its defects, while another is to change the structure into a flexible system by absorbing the fluctuations as given. This paper takes the latter approach, and tries to hypothesize a brand new system suited for the digital age through a creative destruction process.
    Download PDF (384K)
  • Eiichiro YOSHIDA, Toshiyuki AMANO, Yukio SATO
    Session ID: 02-07-25
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, the art museum support by digital archives attracts attention. However, there is no research which treated digital archives and image processing for restoration systematically. In this paper, in order to support actual restoration by virtual restoration, it discussed systematically to digital archives, extraction of damage, and virtual restoration. First, we extracted the missing part by digitizing form for oil painting with flaking. Next, the interpolation of image processing corresponding to the actual restoration was discussed, and virtual restoration was performed. Furthermore, addition of texture was examined in order to improve an interpolation result.
    Download PDF (708K)
  • Kazuto MIYASHITA, Yasushi HIRANO, Jun-ichi HASEGAWA, Junichiro TORIWAK ...
    Session ID: 02-07-26
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we propose a method to recognize inflammatory nodules from thin slice chest CT images. These nodules can be visually decided their benignancy using medical observations without biopsy or follow-up. These medical observations were quantified to be used in CAD (Computer Aided Diagnosis) by six features: nodules features such as ratio of linear part in the bounding, size, shape complexity, the standard deviation of CT values inside it, two features between nodules and chest walls (Distance and existence of interlobular septum). In order to recognize inflammatory nodules, we use the logistic regression model. The method was applied to eighty-three small and solid nodules including forty-four inflammatory nodules. The accuracy of the recognition was seventy-eight percent. Furthermore, we tested hypothesis for the significance of the features. It was confirmed that the two features (Size and standard deviation of CT values) are significant.
    Download PDF (448K)
  • Hirotaka MATSUSHIMA, Tsuyoshi YAMAKAWA, Tsuyoshi YAMASHIRO, Minoru TOG ...
    Session ID: 02-07-27
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    There are various techniques to decide the threshold for binarization of the image. But in the image which was taken under the bad condition, selection of threshold is very difficult. So we proposed the threshold equation to decide the threshold. In this dissertation threshold equation was derived by the correlation of the optimum threshold values and the average density of the background part which illuminated uniformly under various lighting conditions. But, with just decision of the optimum threshold is not obtained the satisfactory result. Then the image was divided, threshold was set to each territory where it divided. As a result, we could obtain the satisfactory result from the decision of appropriate division and optimum threshold value. Moreover, comparision with the other techniques was able to be performed and validity of our technique was checked.
    Download PDF (453K)
  • Its overview and future trend
    Fumitaka ONO
    Session ID: 02-07-28
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    International standard on still image coding defined in ISO/IEC JTC 1/SC29/WG1 is reviewed putting emphasis on the technology trend and the current stage of standardization by classifying the standard into bi-level image coding, multi-level still image coding and entropy coding, including the issues on Intellectual Property Right. Then future trend of related technology including that of standardization is also studied.
    Download PDF (664K)
  • Canon Patent Activities on Global Standard
    Toshihiko NAKAZAWA
    Session ID: 02-07-29
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Last November, NHK (Nihon Housou Kyoukai) presented a part of Canon patent activities by the program "Project X". Canon has been experienced various patent activities on global standards, so that I introduce several examples of those activities.
    Download PDF (61K)
  • Akihiko IKETANI, Noboru NAKAJIMA, Ryuichi HIRAIKE, Keiji YAMADA
    Session ID: 02-07-30
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper proposes a method for tracking multiple non-rigid objects, based on Backward Region Classification. Non-rigid objects are composed of sub-regions with individual motion. Taking this into consideration, we first find for each sub-region the object which it comes from (Backward Region Classification). Grouping the sub-regions belonging to the same objects gives the current object regions. Previous methods based on Backward Region Classification, where the major goal is to track a single non-rigid object, classify each sub-region either to the object or to the background. The proposed method, in contrast, is capable of tracking multiple non-rigid objects.
    Download PDF (485K)
  • Tomonari NOMOTO, Tsuyoshi YAMASIRO
    Session ID: 02-07-31
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    By the spread of a personal computer and the Internet, multimedia environment such as music and a picture, became familiar. Moreover, the research of the moving image processing is also briskly done in recent years. At present, to hide the wanted area of the image that is the telecast on the television is carried out manually. The mosaic for the protecting area cannot be applied with a real time while taking with a camera. In this study, We tried to hide area automatically, by using the color infomations, with real time.
    Download PDF (418K)
  • Andres FUKUHARA, Tsuyoshi YAMASHIRO
    Session ID: 02-07-32
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently, the implementation of the automatic moving object tracking system, in various circumstances, is required to protect us from the increase of crime in our society. Background different method and Frame different method are used to extract and track moving objects from motion image sequences. But when noises occurs by the change of illumination, the extraction of the objects become unstable. Therefore, the tracking of the moving objects become difficult. In this study, a real time tracking method based on background different method, that is not affected by the noises and operate, is proposed.
    Download PDF (399K)
  • Satoshi MOROZUMI, Nobuhiko MUKAI, Makoto KOSUGI
    Session ID: 02-07-33
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    It is generally time-consuming to search for useful scenes in a long time video. For efficient retrieval, it is important to make indices of the video in advance. In this paper, the architecture of automatic indexing and summary generation for baseball games is described. At first, the video images are divided into every pitch using the regular camera work. Next, telop information is recognized and an index is added based on the transition of the information. Moreover, pitching action is recognized by a pitcher's behavior and the summary image that collects only pitching scenes is generated. Furthermore, another summary for only score scenes is also generated.
    Download PDF (313K)
  • Color shift detection of color portions
    Keitaro Kawai, Norimasa Hikosaka, Makoto Ohoki, Sumihisa Hashiguchi, M ...
    Session ID: 02-07-34
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A method is proposed for detection and correction of reading errors in the images scanned by a sheet-through scanner. In portions with high contrast the misalignments among the R, G and B edges are detected as errors. The corrections are performed by shifting both the right edge and the left edge to the position of the center edge.
    Download PDF (231K)
  • Correction due to the hue of errors
    Norimasa Hikosaka, Keitaro Kawai, Makoto Ohoki, Sumihisa Hashiguchi, M ...
    Session ID: 02-07-35
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A method is proposed to correct reading errors of monochrome images. Errors are detected by searching for colored portions, and the correction are performed on the basis of the hue and the width of the colored portions.
    Download PDF (366K)
  • Misato MOCHIDUKI, Hiromitu NISHIMURA, Takehiko TOMIKAWA
    Session ID: 02-07-36
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Man, normally, takes in the impression acquired from a picture through vision. Probably, the feature extraction of the picture by computer is indispensable in order for a visually impaired person to take in the same impression. In the past, there was the report of "1/f fluctuation" so to speak, which made the spectrum inclination from the 1-dimensional Fourier Transform of a picture. However, as for considering the Fourier Transform itself as analysis including the local feature of the picture, the "frequency information" which man receives from a picture had the difficult understandings, in order to be dependent on a spatial location. This authors considered the meaning of the spectrum information at the time of applying to a picture, which has in consideration of the point that the information acquired from Wavelet Transform is the function of time. As a result of performing Wavelet Transform by one dimension for a color picture as the first step, frequency information was greatly dependent on the position information on a picture. Moreover, it also turns out that an interest frequency change in saturation, hue, and value of a color pictures are shown. Although it did not result to grasped the tendency which can generalize a picture, it is believed that there exist the correlation with human's vision and the frequency characteristics.
    Download PDF (361K)
  • Inspection of an optimal retrieval parameter
    akihiko WATABE, jun USUKI, kazumi KOMIYA
    Session ID: 02-07-37
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a new feature description method of video using statistics processing for motion vector. A new method has advantages of fast retrieval time for pick up any desired specific scenes. As the picture of the retrieval object baseball scene studied and optimal parameter were clarified to extract the pitching scenes.
    Download PDF (330K)
  • Yoshinori Itoh, Kazumi Komiya, Tetsuri Inoue
    Session ID: 02-07-38
    Published: March 10, 2003
    Released: August 29, 2003
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, a new collaborative working space model networked IPT(Immersive Projection Technology)and PC was proposed. At first stage, appearance performance and processor speed were studied using an experimented model of virtual play catch. From these performance test possibility of, the system was confirmed.
    Download PDF (358K)
feedback
Top