基于空间模糊化表示的图形相似性测度

根据人的视觉感知的模糊性,提出了一种新的图形相似性测度的计算方法。首先将图形在空间上进行模糊化处理,将一个确定的图形模糊化为二维平面上的一个模糊子集,将两个图形间的相似性测度问题转化为图形元素在图形模糊集中的隶属度问题,从而可以定量地表示相似性的程度。最后给出了用模糊相似性测度实现图像检索的例子。实验表明,该测度算法的结果和人的主观感受是基本一致的。

Similarity measure between shapes based on spatial fuzzy representation

A novel method of measuring the similarity between two shapes is presented according to the characteristic of fuzziness in human visual perception. Firstly the spatial fuzziness of the shape is made, so that the shape is transformed to a fuzzy set in 2-D plane. Then, the measure of similarity between two shapes is translated to the membership of one shape item in a shape fuzzy set, so that the similarity extent can be computed. An example of image retrieval based on the fuzzy similarity measure is also given, and the experiment indicates that the result of the fuzzy similarity algorithm is consistent with the human visual perception approximately.