Kazumi Yamashita

Three-dimensional motion analysis and structure recovering by multistage Hough transform

A new approach for the detection of motion of three-dimensional rigid bodies from two- dimensional images is presented. The approach is based on two main stages. In the first stage, the positions and velocities of feature points are detected from two-dimensional images. In the second stage, the rotation and the translation velocity of each body is detected from the positions and velocities of the set of feature points. We employ the Hough transform method in both stages. We describe the details of the second stage and the method of computation reduction in Hough transform. The effectiveness of our method is confirmed through computer experiment.

A study of the syllable oriented recognition of continuous speech

For very large vocabulary recognition systems, there are suitable smaller recognition units than words because the use of these units has the advantage over conventional word-matching recognition that the recognition system has no limitation for the vocabulary size and can handle large vocabularies with far less recognition computation. The Japanese syllable corresponds to a character called Kana and the syllables are limited to about a hundred. Therefore, the syllable is expected to be suitable as a recognition unit in Japanese. As the size of the vocabulary grows, it becomes imperative that such a recognition system should, as one might expect, operate with higher syllablerecognition accuracy. In order to design a largevocabulary, syllable-oriented recognition system, it is very important to investigate the statistical properties of the recognition system and obtain in advance the c, riterion for syllable accuracy required for the :recognition system. In this paper, we describe a syllable-oriented recognition system of continuous speech in Japanese. It is mainly based on a segmentation strategy and a syllable-template-matching recognition approach. The statistical properties of word and phrase recognitions are examined using a simulator of monosyllable recognition. We summarize the results and discuss the implications for syllable-oriented recognition strategies. Recognition system

High-efficiency image-coding method using vector quantization and block coding

This paper proposes a high-efficiency image-coding method which is a two-step coding using a vector quantization and an adaptive block coding. This method is equivalent to separate quantizations of low- and high-frequency components of an image signal, and was applied to the luminance signal and monochrome image obtained from a color image. The results show that this method is superior to conventional vector quantization and adaptive block-coding method. The method was also applied successfully to a color image by converting its R, G and B signals to the Y, u and v signals, and by applying the two-step coding method to its Y signal (luminance) and the u and v signals (chromatic ones).

Array processing of wide band signals with computation reduction in coherent and incoherent environment

A high resolution iterative algorithm for estimating the direction-of-arrival (DOA) of multiple wideband sources is proposed. For equally spaced array structures, two unitary transform based approaches are proposed in the frequency domain for signal subspace processing in both coherent multipath and incoherent environments. Given prior knowledge of the initial estimate of the DOA, with proper spatial prefiltering to separate multiple groups of closely spaced sources, the proposed algorithm is shown to have a high resolution capability even in coherent multipath environments without reducing the angular resolution as compared to the use of subarrays. The performance of the algorithm is demonstrated using simulated data.<<ETX>>

Real-time video signal processing by generalized DDA and control memories: three-dimensional rotation and mapping

A new method for video signal processing is described in this paper. The purpose is real-time image transformations at low cost, low power, and small size hardware. This is impossible without special hardware. Here generalized digital differential analyzer (DDA) and control memory (CM) play a very important role. Then indentation, which is called jaggy, is caused on the boundary of a background and a foreground accompanied with the processing. Jaggy does not occur inside the transformed image because of adopting linear interpretation. But it does occur inherently on the boundary of the background and the transformed images. It causes deterioration of image quality, and must be avoided. There are two well-know ways to improve image quality, blurring and supersampling. The former does not have much effect, and the latter has the much higher cost of computing. As a means of settling such a trouble, a method is proposed, which searches for positions that may arise jaggy and smooths such points. Computer simulations based on the real data from VTR, one scene of a movie, are presented to demonstrate our proposed scheme using DDA and CMs and to confirm the effectiveness on various transformations.

Transformation of orthogonal coordinates into polar coordinates using DDA‐computation of the angle of rotation of an object

This paper describes the algorithm which directly processes the gray-level image to determine the rotational angle. A complex computation is required to transform the orthogonal coordinates into the polar coordinates. In this paper, a transformation algorithm is proposed using DDA, which is a high-speed algorithm to generate the digital segment. To evaluate the usefulness of the proposed algorithm, a computer experiment was performed using several input images from a TV camera. The mean error of the rotational angle was ±0.8 deg, and the maximum error was ±3.9 deg, which is quite satisfactory.