基于团块的特征提取

BLOB-BASED FEATURE EXTRACTION

本文提出了一种新的基于四块的图像处理思想，描述了在团决的链码表示的基础上提取其点、线、面特征的算法．与以往的图像处理方式相比，新方式更接近人眼观察景物的方式，并可在图像处理的早期引入知识以控制搜索空间，减少计算量，加快运算速度．

This paper presents a new blob based image processing idea, and describes the algorithms of feature extraction based on the chain code of a blob. Compared with the traditional methods, the new idea conforms better with the way of human's observation of scenes. The algorithm presented here introduces knowledge in the early stage of the image processing period, so as to construct the searching space, reduce computation amount, and speed up the processing considerably.