The region on which a picture is defined is subdivided into nonuniform sections (blocks) according to its spatial frequency. The original luminance values in a block are approximated by two triangular plane patches constructed by the luminance values at the four corners of the block. Luminance discontinuities appeared in the boundaries of blocks are corrected by using the luminance values of smaller blocks. An experiment on a digital picture make it clear that the details of the picture are reproduced by the correction of the luminance discontinuities. The picture represented in YCrCb space is encoded efficiently.

View full abstract

In transform image coding such as DCT, an image to be encoded is usually partitioned into square-blocks. As this segmentation of the image is carried out independent of its contents, such coding schemes have some problems concerning coding noise which deteriorates reproduced image quality, e.g. blocking-effect and mosquito noise. In this paper. we propose a new coding scheme in which all square blocks are transfigured into variable shape blocks so that their boundaries can fit the contours of the image. And luminance component in each block is encoded using KLT basis-functions derived from the covariance matrix with respect to its block shape. Computer simulation indicates that this coding scheme provides not only higher coding performance but also better reproduced image quality compared to square-block based coding such as JPEG.

View full abstract

This report describes the implementation of ATM adaptation layer for broadcast video service in ATM network, which is based on block coding and variable bit rate coding and uses the coding block as the ATM adaptation processing block. We propose a multiplexing method based on SAR protocol type 2 for video and audio, a cell loss detection and compensation method using parity cell and sequence number composed of block and cell number, and a media clock recovery method using reference timing of block information. Furthermore we describe our new ATM adaptation board using these methods.

View full abstract

Structural representations are effective in handling moving pictures by computer. In this paper, we show the idea of a structural representation and automated scripts to generate this representation. The method of generating scripts consists of extracting object from moving pictures, calculating these features and generating the scripts from these features. At last, we show a result of making computed structural representations from moving pictures.

View full abstract

In this paper we are proposing a similar video detection method in human walking samples by using two following features. The first feature is a walking speed that is rate of foot moving gained by means of fourier transfer procedure. The second one is a walking stride that is the length of a walking step gained as an average of maximum values in each cycles. We confirm that each detected video is the same one selected subjectively by human. We intend to construct the video wandering (retrieval) tool that uses the detection method.

View full abstract

Holographic pattern measuring system is a combined system of a holography and a graphic image processing technique. A distribution of microscopical displacement can be measured, and it can be analyzed as the three-dimensional graphic image. However, in the system, the interferometry data have been treated handly, so both the error and the time loss were inevitable. In this Study, the digital image signal processing system was developed, and the system was applied to detect fringes automatically from interferometry image. AS the results, more accurate pattern data could be clearly and speedily detected by using peek-detected processing method.

View full abstract