Through mutual technology transfer between consumer and professional video recorders, the last 20 years has witnessed a rapid evolution from analog to digital recording. Each new digital videotape recorder (VTR) uses different channel coding. This implies that improving the recording density involves the development of new channel-coding schemes together with evolutionary magnetic tapes and heads. As a result, professional digital VTR's offer the best features for video recording. This paper reviews state-of-the-art magnetic recording devices, signal processing techniques for digital recording. It also describes the specifications for home-use digital recorders for current television systems, and discusses disk recording technology in the future.>

An extended definition TV (EDTV) system which is fully compatible with the existing standards is proposed. Full compatibility implies receiver and transmission compatibility: both existing and EDTV receivers can receive existing and EDTV signals which have the same bandwidth, without adaptors. In order to insert high resolution information, vacant frequency bands are found in those areas in the first and third quadrants that conjugate with the modulated color components in the second and fourth quadrants when the signal is analyzed in the temporal-vertical frequency domain. Resolution can be improved both horizontally, by inserted high resolution information, and vertically, by progressive scanning and line interpolation. An alternate scheme is also proposed. These schemes are applicable to CATV, CCTV, video storages, broadcasting, and direct broadcasting satellites, without any modification of existing facilities.

A new coding scheme for dithered halftone pictures, called `pattern length' coding, is proposed. This scheme is based on the fact that, in dithered pictures, some special horizontal bit-patterns frequently take place. Counting the pattern strings provides coding efficiency that is as good as any other scheme presented so far. In addition, this scheme is suitable for pictures which have both ordinary document areas and halftone areas, thus keeping better compatibility with standard facsimile coding schemes.

A very high-speed freeze-picture transmission system via telephone lines is realized by sending only the changed areas from the previous picture. Transmission speed is improved in reverse proportion to the changed areas compared with conventional systems. Increase in circuit size is nominal. The new system is also much simpler than interframe codecs because it does not employ complicated coding mode controls, addressing, or buffering. Block size of the changed areas is selected to satisfy both fast transmission speed and circuit simplicity. High-speed transmission compatible with high picture quality is achieved by extraction of changed blocks using both the average value of the picture signal difference in each block and the maximum value. Protection of picture data against transmission errors is achieved using two synchronizing data groups among tripled addressing data groups. Reduction of impairments due to errors in DPCM picture data using backward decoding is also proposed and adopted.

With a facsimile bi-level quantizer, notches often occur along the white and black borders of a document when the output of a photosensor goes up and down around a threshold. The objects of this work are, first, to present a notchless bi-level quantizer, and secondly, to improve coding efficiency by using the quantizer. It is pointed out that quantization is comprised of many important aspects from aesthetic as well as coding efficiency standpoints, and that significant improvement is achieved in coding efficiency through the use of the notchless bi-level quantizer.

A new run-length decoding method for facsimile signal transmission is described. In this method, if an error code is detected in a received signal, the effect is localized in the runs corresponding to the error instead of the whole line to obtain better picture quality. The error immunity in this method can be used to shorten the transmission time.

A study is made of differential PCM (D-PCM), which is a relatively simple but efficient coding system for TV signals. The purpose of this paper is to optimize the D-PCM system with special consideration of the visual property, that is, to obtain the optimum prediction coefficient and the optimum companding characteristics with consideration of the contract sensitivity used in D-PCM. It is found that when the visual property is considered, the optimum prediction coefficientaapproaches 1.0 or complete integration (i.e., integration with an infinite leak time constant), although when the visual property is not considered, the optimum prediction coefficientais the ratio of autocovariances, as we know from previous presentations.

Various image coding schemes have been studied for digital transmission of videophone signals. The Hadamard transform, which is now studied for the transmission of pictures such as those from satellites, has been considered too complicated for public use, though the characteristics such as the ratio of bit-rate reduction are more desirable than those of differential pulse-code modulation (DPCM). We have found a very simple scheme of the transform, where digitized videophone signals are transformed to Hadamard components all digitally just byndigital adders and some shift registers for 2nth-order transform. For example, three adders are necessary for eighth-order transform. It is extendable to two-dimensional transform with ease. We have made an experimental model running in real time. Experiments and theoretical calculation have shown that 3 bits/sample are required for good picture quality in the case of two-dimensional(4 \times 2)th transform and 0.5 bits more for one-dimensional eighthorder transform.

Word synchronization in PCM of TV signals is more important than that of audio signals, since often a single channel of TV signals is transmitted independently. For this purpose, new word synchronization is proposed, where the statistical property of TV signals is utilized instead of adding a bit for synchronization. The property is that a more significant bit in PCM does not change word by word as frequently as a less significant bit, because of the correlation between adjacent picture elements, i.e., sampled values or words. Mathematical calculation shows that the average time between misframes is as long as 104years and the average recovery time is less than several lines when a 4-bit reversible counter for synchronization is used in transmitting 500-kHz narrow-band TV-phone signals.