Vector quantization techniques have been widely investigated because of the remarkable efficiency of vector quantization. This paper describes a hybrid coding method consisting of vector and scalar quantization. Coefficient blocks, obtained by orthogonal transformation of pixel blocks, are first vector quantized, after which residua of the vector quantizing are then scalar quantized. A classification method using vector quantization criteria which improves scalar quantization efficiency, is also presented.
A new image analysis method for metallographical image is described. In this method, the spectrum of every line scanning signal of the subject image is calculated using Maximum Entropy Method, and, from these spectra data, the histogram of the appearance of every frequency in the subject image is generated. The characteristics of metallographical image can be analysed stochastically from this histogram. From the results of the computer simulations, it is shown that this method is useful especially for fractographical image analysis.
This paper describes a color image quantization scheme for presentation of color images using small number of colors. This scheme consists of three steps (1)partition of a color space into many regions, (2)merging of the regions until below prescribed number, (3)selection of a color representing the region. This paper also presents a scheme for presentation of color images using fixed set of colors. The representative colors are first classified according to their hue and are converted into the fixed set of colors according to their lightness.
A hierarchical chain coding scheme is proposed for constructing large-scale line-drawing data base systems. The method transforms a line drawing image into a pyramid structure by uniting two links in each layer, and has the following features : 1) Arbitrary portions of a line drawing image can be retrieved in arbitrary resolution with a factor of 2. 2) Related portions of link code can be accessed in a short time. 3)Partitioning of link information into layers does not cause a significant increase in overhead code amount. In an experiment, a complicated coastline image was encoded in almost the same efficiency of Freeman's code, 3 bits/link.
The thin line preservation method (TP method) is considered to be one of the best picture size conversion method for binary pictures with respect to the converted picture quality. This report describes the algorithm and experimental results of the extended TP method (ETP method) which refers to more original picture elements than that of TP method to improve converted picture quality when conversion ratios are less than one half.
A public facsimile communication network developed for facsimile communication media by NTT has been offering store and forward type communication processing services. Facsimile communication services offered by this network are classified into two groups, message transmission services and information retrieval services. This paper describes about user interface classifications these new communicaton services and proposes some new user interface techniques.
This paper describes the measurments and its results of transmission characteristics for 2 types of facsimile broadcasting system which use the data channel of the direct broadcasting satellite (DBS). The first system uses an analog facsimile signals. These signals multiplex on the data channel in PCM coded form. The second system uses a MH coded facsimile signals. These signals multiplex on the data channel in packet form. In the laboratory test, the transmission characteristics of these facsimile signals were measured using the experimental systems for DBS. In the first system, the margin of CN ratio is 0.3dB and the objective value of facsimile picture is 3 at CN ratio of 6 DB. In the second system, the margin of CN ratio is 1.7dB and the objecive value of the facsimile picture is 5 at CN ratio of 6 DB.
A facsimile broadcasting system multiplexed with the television sound channel have been studied in Japan. In this system the facsimile signal is FM-FM modulated using the 2nd subcarrier and the facsimile control signal composed of program and page number, etc, is transmitted in NRZ code form, on the other hand the facsimile picture is transmitted in analog form. This paper describes the analysis and experiments of the impulsive noise disturbance on the reception of the facsimile control code signal. The bit error rate of the code is investigated by the analysis of the FM-FM demodulation. While, using the indoor transmitter and receiver the bit error rate and the frame error rate of the code is measured. Through the investigation it is cleared that the bit rate of 8000 bps can be used and (272,190) majority-logic-decodable-difference-cet cyclic code is effective for the frame error correction.