Space-Time Vector Quantizing of Color Imagery

由VLSI实现的图像的数字传输被认为是一个非常重要的领域，因为在可接受的误差水平下，传输的信息速率可能会降低。到目前为止，大部分研究都致力于静止图像或帧内编码。其他工作已经针对帧间编码或运动编码任务。在本研究中，追求帧内(静止)图像编码和帧内(运动)图像编码两者，其目标是进一步降低比目前已知技术更低的比特率。这项研究涉及到矢量量化原理，该原理将扩展到彩色图像和时间编码。目标是使目标比特率大大低于当前水平。第一个主题涉及将最近在帧内编码方面的工作扩展到彩色图像和更低的比特率。还提出了通过对帧差信号进行矢量量化来对运动图像进行编码，以及一种简单的分类方案，以便以标准视频会议编码器目前存在的大约三分之一的速率来处理彩色图像。

Digital transmission of images implemented by VLSI is considered a very important area because of the possible reduction in the transmitted information rate subject to an acceptable level of error. Much of the research to the present time has been devoted to still-picture, or intraframe coding. Other work has been directed towards the interframe coding, or motion coding, task. In this investigation, both intraframe (still) image coding and intraframe (motion) image coding, with an objective of further bit rate reduction below presently known techniques, is pursued. This research involves vector quantization principles which are to be extended to color imagery and to temporal coding. The goal is to make the objective bit rate substantially lower than present levels. The first topic involves extending recent work in intraframe coding to color imagery and to lower bit rates. Coding of motion imagery by vector quantizing of the frame difference signals is also proposed, along with a simple classification scheme to process color imagery at rates of approximately one-third of the rate that now exists for standard videoconferencing coders.