The video coding is one of the most important technologies to realize the economical visual communication and has been widely studied. Especially, in the field of video teleconference and videophone, new coding technologies have been developed in order to reduce transmission rate drastically. In this paper, some of key technologies for low bit rate coding are surveyed, and outline of video codec for teleconference and videophone are also presented.
A wide variety of low bit rate video codecs for teleconferencing and picture telephony have been proposed. We have built a 32 kbps to 1.5 Mbps color video codec using an adaptive hybrid coding method. This coding method consists of a discrete cosine transform, a new two stage frequency domain motion detection method, and adaptive prediction using 4 different predictors. In this paper we explain in detail this coding method and present examples of coding performance obtained from both computer simulation and hardware evaluation.
This paper proposed an algorithm to improve the quality of restored image step by step. The object images used here consist of 704×576 pixels with luminance signal Y and chrominance signals Cb and Cr. The first stage encoder approximates the original images by flat and curved surfaces while keeping edge addresses. The second stage encoder is used to process the error images by Block Truncation Coding. The error images are obtained by subtracting the decoded image from the original image. Computer simulation results show that the usefulness of this algorithm from the view point of subjectively judged quality of decoded images.
Arithmetic Coding is a very effective means to achieve high coding performance in DPCM for TV signals. However, its application for TV signals is very difficult because high speed real time processing is necessary. For the purpose of considering hardware realization, in this paper, we propose a method of hardware implimentation of Arithmetic Coding. Its main feature is the ingenious replacement of the many shifts and addtion processes by Read Only Memories and Data selectors. As a result, it is clear that Arithmetic Coding is fully applicable to TV coding by using the proposed method.
A universal video signal processing simulator has been developed to promote various kinds of image processing studies economically and efficiently. This system consists of a large-capacity ssmiconductor memory device (totally about 3Gbits) and a high-performance minicomputer. The system processes any type of signals such as NTSC, HDTV, noninterlaced TV signal and so on. Since the system has a module structure, it is easy to expand the memory capacity and to improve the video signal transfer speed.
This papaer describes a new feature extraction method from color images. The chrominance signals are considered to be a vector field of two dimension. In the vector field theory, any vector can be divided into two components, lamellar and vortex. Lamellar component describes potential field and vortex describes the rotation of the field. For the color image, lamellar component represents the global structure of the image and vortex component represents the boundary information. Using these properties, we propose the new coding scheme which is similar to plateau coding.
This paper describes the automatic index extraction algorithm for document database. First, the layout structure of the input documents is obtained in the storing process. And then the index information is extracted by matching the input layout structure with the document format information. This algorithm consists of the following three steps : (i) To recognize the category of input documents, (ii) To distinguish the index regions, (iii) To store the index information. Finally, the computer simulation results for the technical papers are presented.