A HDTV still picture system has been developed for an HDTV program production. This system consists of character generation video file, painting system and so on. In the video file, after JPEG signal compression, still pictures are stored onto a rewritable 5-inch optical disk. New equipments in this system are HDTV character generator and video file for studios. The character generator, which has ten kinds of Japanese character font sets, has been developed to meet the demand of production side people on designing pictures. The new still video file has been improved for studios on reproducing a scroll with a high transmission speed and allowing to fetch, stick or delete images defined rectangle area in the picture.

View full abstract

We has developed a novel digitized softness chrmakey system which produces extremely natural images for HDTV. The main feature of this system is eliminating, smear on thecontour of keyer imazes. An additional key-signal, incorporated with the conventional soft key-signal, is introduced into this system. It achieves a soft matching of the contour of the foreground object,which used to cause smear blurred by the keyed color.

View full abstract

A region extraction method is an important technology for video synthesis. We have already developed an extraction method which can automatically extract a necessary object from an arbitrary video after roughly specifying it. We have just developed a Hi-Vision object extraction system base on this method. In extraction, an operator is only required to specify an object in first two fields. After that the system automatically extract the object from a video getting a field image from a VTR, extracting an object in a filed and storing key data to a disk in all fields of a specified scene. In synthesis, background and foreground video are composed in real time using key video stored in a disk . There are other features, processes in high quality by digital interface, simple specification of an object by paint function, fast and exact extraction processes by specially developed hardware and so on.

View full abstract

Prototype model of 32Mbps Hi-Vision CODEC(HERCULES-32M) has been developed for the purpose of industrial applications such as monitoring electric power facilities, and construction site, etc. Hi-Vision video tape pictures of the power facilities were made to evaluate the Hi-vision codec's picture quality. Then, the codecs were connected to the digital network operating at 32Mbps, and hydro-power station monitoring was imitated. As a result of these tests, effectiveness of Hi-vision remote monitoring using 32Mbps codec was confirmed.

View full abstract

We performed a field trial of digital TV transmissions by the MUCCS45 (45Mbps MUlti-Channel TV Coding System) and analog FM TV transmissions, as well as digital・analog mixed transmissions on INTELSAT satellite. In this papaer, we report test results about transmission performances and quality assessments based on a subjective picture quality evaluation test.

View full abstract

This paper proposes new ways to look at orthogonal transforms developed for image-data compression. The impulse response is introduced to evaluate low-pass filtering resulting from the truncation of transform coefficients in the first stage of data compression. Fourier analysis of the basis vectors is proposed for testing noise shaping capabilities, where effects of quantization noise are assumed to be monitored not in the transform sequency domain but in the traditional Fourier frequency domain. Sample results involving the DCT and SCT are included to show that they are quite varied in the new light despite that they look alike when compared in terms of traditional performance measures such as residual correlation and energy packing efficiency.

View full abstract

Recently, DCT is the most popular in image coding using orthogonal transform. Because its performance is closely to that of Karhunen Loeve transform, which is known to be statistically optimal. But it, is under very high ρ condition. If the ρ is low, other orthogonal transforms may have the advantage of DCT in some points of its performance. This paper presents the selection of orthogonal transform for Interframe Prediction Residues Coding using statistical signal models.

View full abstract

This paper proposes a new coding technique for video signal transmission. It encodes interframe prediction residues with edge patterns in the wavelet transform domains. It can achieve high compression by a suitable allocation of bits to each wavelet signal. Efficacy of the proposed method is confirmed by simulation.

View full abstract

This paper proposes a new method to design a vector quantizer for wavelet-transformed images. The proposed method uses random vectors as the training set,while in the conventional method the training set for the quantizer is derived from different images to be wavelet-transformed. Computer simulation is inclued to show that the coding efficiency with the proposed method is the same as that of the conventional method.

View full abstract

Transform image coder produces images with objectionable degradation such as blockiness and edge blur at low bitrate. To prevent these degradation, we propose a scheme based on adaptive bit assigning to classified several subblocks. Subblocks are classified into four groups according to their activity measured by variance of pixel value including the subblock. Experimental results show that reasonably reconstructed image quality is obtained. Furthermore hybrid method with edge-pattern image coding is studied.

View full abstract

The coding scheme of interframe DPCM mixed With an orthogonal transform hybrid coding such as MC DCT is currently a dominant scheme. In this scheme, a temporal correlation varies according to a transform coefficient, further, a predictive accuracy also varies because of quantization noise being fed back. Then. in this paper, both a predictive method (predictive coeffcients) and a quantizer control are commonly handled, and these two factors are theoretically optimized to give the highest picture quality to viewers. Computer simulation results based on this analysis show siginificant coding gain can be obtained comparing with the predictive・quantization scheme used in MPEG.

View full abstract

We studied on modification of the image concealment method. Embedded binary data are blocked by variable length, in our method. The simulation results show that the embedded data amounts are 22 percent to 45 percent higher than the conventional method. We ,moreover, considered about the method to embed coded musical score data.

View full abstract

Effective Tele-Monitoring Systems (TMS) are essential if objects are to be remotely monitored at reasonable cost. We originally proposed Catch Vision to answer this need. This paper extends Catch Vision by proposing the Adaptive Sampling Method (ASM). The image sampling rate is coupled to the velocity of the watched object. The subject was captured at the maximum operating speed of camera. Frames are held in memory and selectively released. Tests were conducted using CG images. It was found that, at the same average bit rate (averaging time of 5~6 s). ASM is superior to the fixed sampling period method.

View full abstract

We have developed a Video Window Processor (VWP) which overlays ITV images in an operational monitor of plant monitoring system, This processor's features are (1)displaying multi-video windows system using fast copy method in frame memory, (2) displaying high quality magnification image using bi-linear interpolation method, (3)displaying 1024 pixels x 860 lines x 60 frames/sec, In this paper, we will describe a high quality color video windows system using the VWP.

View full abstract