We examine image retrieval from form without using keywords on the subject that is unknown name. The subject of our retrieval is images of earthen. The feature of shape described parameters of Superquadrics. Superquadrics can describe a wide variety of shapes with just a few simple parameters, and recognize shapes easily. The parameters extraction method makes use of closely resembling least-squares of Superqudrics on the shape of earthen. We could get effective as a result of extending Superquadrics and adding a new parameter that describes shape of the upper part of an earthen. Experimental results of retrieval show that the proposed method is effective.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

The Virtual Printing System simulates real woodblock printing: generating a woodblock, spreading inks on the woodblock, putting a paper sheet on the woodblock, and rubbing the paper sheet with a Japanese squeegee BAREN in a virtual space. We study digital archives of Japanese prints Ukiyo-e based on the virtual printing method. We developed new virtual ink to create Ukiyo-e with the virtual printing.

View full abstract

In this article, digital imaging system for old documents was discussed. The requirements for the system were pointed out, and one of them, improvement of legibility, was examined. In the old documents, transparent characters from reverse side of paper degrade the legibility of characters on front side. Therefore, in the improvement of legibility, frequency filters were used to segment characters on front or reverse side, and hi-pass and low-pass filters which have directional intensity corresponding to the human visual system were operated to improve the legibility. As the results, the characters on front side were segmented well, but it was not enough for the characters on reverse side.

View full abstract

Recently, the digitalization material such as historical document and the valuable books is being carried out by many organizations aiming to develop electronic libraries and museums. Such digitalization efforts often fail to adequately address important issues concerning document preservation and digital content delivery. Moreover, even in the case of providing access via WWW, these digital presentations provide merely a photographic record of the original materials. In this paper, we describe the process for digitalization of 4 documents including “The Document of Japanese Economic History” taking into consideration aspects such as document preservation and optimal content delivery, Further, We discuss various aspects relating to the development a “Virtual library” which would provide an interactive environment for exchanging opinions and comments about these documents. A walk through and “virtual bookshelf” feature are newly implemented for the document retrieval. Several user-interfaces related to the “virtual library” are presented and an information retrieval based on Z39.50 standard using Open Text5.1 is also discussed.

View full abstract

The examination result about the framework for carrying out crossing reference of the document information, which two or more museums exhibit, which inquired in the Institute of Image Electronics Engineers of Japan VMA study group “the document type definition SG of a museum and an art museum document” (DTD-SG) is reported. A framework consists of I) information structure level, II) information description level, and an III) information navigation level, and performs a share and mutuat conversion of the structure (template) for describing information on a level I. On a level II, a share and mutual conversion of information of the vocabulary of the contents of description are performed. And on a level III, the share and mutual conversion of related information (link) out of line for navigation are performed.

View full abstract

This is a trial design of the super intelligent multiple search and retrieve information from the every museum in Japan. With it, we can find any information of exhibited metadata from them over the internet using ANSI/NISO Z39.50 protocol and converting to Dublin Core metadata structure. We can also view the digital pictures and/or explanations of the objects with linking from particular metadata to the database of the actual museum. To make this experiment, we used each metadata of Tokyo National Museum, Kyoto National Museum, Nara National Museum and National Museum of Japanese History with many thanks.

View full abstract

NewsML is the standard format designed for newspaper companies and news agencies to interchange and manage news data. NewsML is published by International Press Telecommunications Council (IPTC) in October 2000. NewsML, the standard based on XML and other appropriate standards and recommendations, provides flexible frameworks to handle arbitrary mixtures of media types, formats, languages and encodings. I will explain “The structure of the NewsML Document Type Definition”, “how to describe vocabularies, which are controlled values of attributes, in handling NewsML documents universally and how to control the vocabularies”, “metadata for easy interchange of data”, and “how to create a NewsML document and interchange it”. While NewsML is originally a standard for news data, it is supposed to be appropriate format to interchange and manage the other type of data.

View full abstract

Optical disk systems will replace VCR systems in broadcasting stations in the future because of their advantageous random access and high endurance features. A 100 Mbps read/write experiment using a laser beam for broadcasts has already been conducted, but 100 Mbps is only a minimum rate. Playback at higher rates is needed for transmitting of collected materials and post production. To address this issue, we have studied ways to achieve a higher transfer rate in playback and nonlinear editing in units comprising a certain number of frames. Performance of current optical disk drive enables two times higher transfer rate in outer area, and also enables nonlinear editing in an 8-frame unit, by using constant-angular-velocity (CAV) rotation method. We conclude that using both recording the outer area and the CAV rotation method is a suitable approach for newsgathering.

View full abstract

Recently, national museums all over the world have established digital archives to preserve their precious collections. Traditional digital archives contain artificially realistic collections to produce a feeling of immersion by the visitors. But this is insufficient for creating new artistic collections in the museums. Therefore, we have been thinking about how to create a novel digital archive for the Chinese Bronzes Collection which is interactive. This paper is a proposal for generating a novel archive by using Computer Graphics (CG) and Virtual Reality (VR) technology. This paper shows how to generating a CG shape’s modeling of Complex Chinese bronze automatically and how to use a SPIDAR system of VR. Especially, visitors using a SPIDAR system to strike one of the CG shapes of the Chinese bronzes, the SPIDAR system immediately begins to perform sounds, which come from collections of the actual bronzes. Music is automatically generated and the volume depends on the force the user uses to strike the bronzes. Using the know-how of this paper, it’s very easy to create many novel digital archives with new art in traditional museums. The finally goal of this paper is to present a novel digital archive generation from a traditional museum and to show new art by using an interactive system.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This paper describes a smart remote sensing system that can detect scattered mine weapons by making use of an infrared ray (IR) camera. The object (mine) detection method proposed used parallel image processing based on time difference or subtraction images from the IR camera. Because of the danger involved, it is desirable for mines to be detected from the most remote endeavoring. In the case that mines are buried in the ground, we can detect them if there is a sufficiently large temperature gradient between the ground and the radiant heat around the mine position. First, a cold water solution is sprayed over the surrounding region to cool the ground. Then we measure the temperature of this surface with an IR camera. In this time, we can obtain time difference images by checking the difference in temperature of the soil and the radiant heat of the buried mine object and the cooling speed. Each subtraction image is processed in real time by using pipelined image processing software implemented on a Linux platform PC.

View full abstract

We introduce an image retrieval method that has the advantage to permit a pattern translation and fluctuation. Our method makes use of 2-dimentional wavelet decomposition (2-D WD) of the query and database images. The coefficients of 2-D WD are used as the multi-resolution image content. The similarity measure between the query and database images is performed after mapping the significant image content onto the original image and estimated by investigating the positions of the mapped content and the operation factor of 2-D WD such as the resolution of decomposition, frequency information of filtering.

View full abstract

With the rapid development of computer and network technology, it is easy to acquire digital images through the Internet or even using a cellular phone equipped with a CCD camera. Consequently, the demand for easily accessible image-database increases. It will be realistic to build an image database that accepts an image as a query-key to search and extract other images, which share some common features with the key image. This research builds a flower image database that receives a flower image taken in the natural environment as an input, and presents some information about the flowers. We propose a database query mechanism based on the feature matching and interactive and iterative dialog between the system and the user for rapid and reliable information retrieval.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In this paper, a method to extract moving objects on a pixel base for MPEG-4 “sprite coding” is proposed. Sprite coding is a form of object coding: it uses a unified panoramic background image derived from a sequence having a camera motion, and a foreground object as video object planes (VOP’s). The proposed algorithm utilizes background difference and watershed transformation to extract the foreground moving objects. We first apply background difference to generate a foreground candidate image. Then, watershed transformation is applied to this candidate image to extract the contours of the foreground objects. Moreover, we utilize the macroblock mask of the foreground objects that is generated by the conventional method to delete the unnecessary edges extracted by watershed transformation. Results given by our proposed method are more than twice as better than that of the conventional method.

View full abstract

Streaming of MPEG video on the network where various bandwidth are intermingled has been getting common. Thus, the research on transcoding, which changes the bit rate of video according to the bandwidth of a transmission line, has paid attention. In the conventional transcoding, degradation of image quality caused by bit rate reduction diffuse accross the whole image. This approach is not so efficient in picture coding system. We focus on realization of more efficient transcoding, which maintain the better image quality in ROI. The experiments we conducted in this paper are foreground extraction from MPEG-2 bitstream and image quality control within transcoder.

View full abstract

[in Japanese]

View full abstract

Since most of color facsimile products seem to be equipped with only JPEG as the encoding method, ringing artifacts are caused in the character image area such as the sender information header. Therefore, I propose new technology that reduces them. The proposed technology is the algorithm that drives out the level of the pixel having potential for becoming a artifact from the range. As a result, the decoded image has little noise, because the decode clips the part stuck out from the range. The technology realizes the lossless transform in the case of being applied for the bi-level image.

View full abstract

The Blue Noise Mask (BNM)is a halftone screen that produces unstructured, visually pleasing halftone images. However, in the construction of the original BNM, a number of constraints were used to limit its characteristics in the spatial and frequency domains. These constraints were not efficient to provide the required adaptability to all gray levels in the screening process. In this paper, we provide a new algorithm to obtain the Pseudo Blue Noise Mask (PBNM) designed by using prime numbers and filtering techniques to reject the unwanted residual low-frequency components. The halftones by PBNM and the usual error diffusion method are compared. According to the psychovisual test, the PBNM gives the result comparable with that of error diffusion technique.

View full abstract

[in Japanese]

View full abstract

In this paper, a new system architecture called “Decoder Downlodable System” is proposed. The purpose of this system is to provide a uniform platform for the multimedia world based on image compressions using characteristics of the contents (Content-oriented Coding). This system aims dynamic decoder downloading by a negotiation between a server and a client for the seamless and minimum delay playback. This negotiation scheme is implemented by Java and CORBA (Common Object Request Broker). Thus, servers and clients in this system can be independent from the OS and the hardware specification. This system targets for TV broadcasting as well as Internet streaming. A method to realize more flexible decoders is proposed in order to achieve minimum delay playback. The proposed method enables that the modules for bit stream parsing are entirely detached from the modules for image decoding. Thus, the flexible decoder can make downloading time shorter and can be extend to multimedia content search system.

View full abstract

A coding scheme specifically designed for animation images is proposed. Taking characteristics of animation images into account, lines and homogeneous color regions are extracted from animation images. Lines and contours of the homogeneous color regions are approximated by straight-line and spline functions. We found that smoothing operations are effective to extract homogeneous regions from background images. Connected Filter is suitable and can keep edges of the original image correctly. As a result, an animation image can be represented by two layers; significant points as the base layer, and DCT blocks for an additional layer. For compensating scheme for the loss caused by smoothing operation, DCT is applied to for the difference between the original images and lines and homogeneous color regions images. High-quality images can be obtained by adding this differential data to the approximated images.

View full abstract

Although the volume rendering (VR) was well used to the three-dimensional medical image, it is difficult to use VR for the surgery simulation and the tele-operation support, because the amount of data reconstructed using VR is enormous. Surface rendering (SR) is effective than VR in there because a vivid 3D medical image can be reconstructed using SR, and moreover the amount of data reconstructed using SR can be reduced very much if a suitable polygon reduction algorithm is used. In this paper, polygon reduction methods were discussed, and a new polygon reduction algorithm was proposed. This reduction algorithm is effective to the triangle surface data, such as the surface data reconstructed by the Marching Cubes and the various improved Marching Cubes algorithms.

View full abstract

We propose a method of classifying the feature points on an image stream for every object, when two or more objects are included, and carrying out the modeling of each object. The big feature of our method is classifying the feature points, without using the object information such as the number of objects and the number of object feature mark. By applying an affine camera model, affine parameters is calculated at all the groups of four feature points, and the feature points is classified by clustering from the similarity of the parameters. And iterative processing is applied in order to make a classification more effective.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In this paper, we propose contour approximation method by Dynamic Programming and Feedback algorithm. Currently, LSI’s processing speed has been dramatically developed, so that Dynamic Programming can be achived by a general-purpose computer. In conventional researches, segmentation of contour is used in order to approximate the line accurately. The number of extracted points, however, greatly increases on some segments. We propose Feedback algorithm to remove redundant points. Data-Error evaluation verifies the effectiveness of the proposed algorithm.

View full abstract

The purpose of this study is to realize much higher pixel density such as 2,000dpi than the current available magnetic printer of 600dpi. As the pixel density increases, it becomes more and more difficult forming a white line between black lines as well as forming a black line with high optical density. The magnetic forces acting on the magnetic toner are calculated with various characteristics of the recording medium and the toner. As a result of calculative simulation some conditions are found to realize a white line as well as a black line with good image quality even at 2,000 dpi.

View full abstract

The development of educational contents, under WBT (Web-based training) technology, shows a constant growth, along with it, an accurate evaluation in terms of understanding of the students and an exhaustive assessment of the teaching methods; and they are related very closely. This paper presents the development of a web-based system for evaluation of students, and also includes a set of powerful features for the processing of this information, to enable the evaluation of the teaching methods. The results will show the tool is easy to use and how it helps the instructor to evaluate the teaching and evaluation methods used.

View full abstract

The proliferation of the Internet and the increase demand for not only voice or simple text communications but rather the combination of voice, video and intensive data transfer has brought about the need to provide broadband access not only to business but also residential and academic institutions. This paper presents the opportunities realized by deploying millimeter wave wireless communications links in the backbone network. The performance in terms of available data rates, BER, channel throughput and other performance parameters of these links are simulated and analyzed. The results indicate wireless millimeter wave communications links are viable solutions for deployment in high-speed backbone networks.

View full abstract

Attention in the recent history of QoS routing, in particular Expedited Forwarding (EF) and Assured Forwarding (AF), has focused on routing or on analyzing the performance of QoS classes separately using approaches not specifically addressing the needs of the classes in question. This study proposes a heuristic scheme on how to route both classes with the emphasis on maximizing admission probability for the EF class. Mainstream algorithms are examined and a case is made for why current algorithms are inadequate and why this scheme, though flawed in its own right, is a step towards finding an optimal and practical solution and hopefully stimulate more research in this issue.

View full abstract

[in Japanese]

View full abstract

The paper analyzes the interference problems between the Bluetooth and the 802.11b IEEE Wireless LAN systems and shows data throughput characteristics.. This paper presents a disposition requirements of these systems and rational analysis of IP level’s throughput.

View full abstract

[in Japanese]

View full abstract

In “the age of convergence of broadcasting and telecommunications, ” advancement of broadband broadcasting will lower the barriers between broadcasting and telecommunications markets. This paper also seeks the possible design of a mechanism that promotes efficient management of broadcasting companies as content providers, measures scale of economy on the operations of satellite broadcasting services that have been already split up, and suggests reorganization of the broadcasting operations led by high-ranking broadcasting companies.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

We describe a method of recognition of human arm in order to construct an object-handling interface of cyberspace. We construct an 3D arm model, and similate the motion of an arm with the model, and build a database of the motion of an arm with the similation image. We recognize the motion of an arm matching input image of real arm with the database.

View full abstract

Digital archiving of audio assets has accelerated in recent years, but in most cases, the original recordings such as SP records are played just once to be converted into digital signals. Signal processing is also done during this process frequently. Then, by fully scanning waxcylinder with laser beams, we’ve tried to the virtual record on computer. This method is non-contact with and non-destructive, it does not damage to records. In waxcylinder, sound signal is recorded in the entire section of sound groove. From the viewpoint, We’ve tried to pick out as much information as possible from the recordings.

View full abstract