规范式边缘模型基运动图象压缩编码方法

Normal Forms EdgeModel Based Compression Coding on Image Sequence

在图象边缘提取与优化的基础上建立了一种新的运动图象信源模型，并通过计算机模拟实现了编解码系统。其基本思想是：根据主观视觉特性，以边缘信息为表达运动图象的空间与时间特征的核心。这种模型基编码方法不需要图象的先验知识，模型建立及方案实现较为简单。针对信源模型，引入基于几何约束的图象匹配方法作为运动分析的基础；并建立了广义运动概念、提出并实现了广义运动分析，丰富了帧间预测编码方法。

A new source model for image sequence based on detecting and optimization image edge established as well as the coder system is simulated by computer. On the basis of subjective vision characters, the basic idea of this paper is: the edge information of an image sequence is the most significant factor which can describe the image's characters both in space and time domain. By this scheme, neither any priori knowledge of an image is needed nor too much complicated techniques must be applied for model's construction or realization. According to the source model, the method of image matching using geometrical constrains is achieved for motion analysis; means of prediction between frames were extended by the concept of general motion introduced; and the algorithm of general motion analysis is completed. The scheme is also characterized by fitting gray value distribution of images using normal polynomials.