图象自相似性测度算法的研究比较

Comparison of the Algorithms on Image Similarity Measure

图象的自相似性测度可作为图象识别和图象压缩的重要特征参量．分形几何图形具有自相似性和递归性，分形维数自然成为有效的自相似性测度．基于目前常用的有效的计算图象分形维数算法，其中包括“毯子”法、“盒子”法、“分数布朗随机场模型”法和各种改进算法，对各种方法的计算量、适用范围进行了研究比较．

The similarity measure of image could be a useful character in image recognition and compression. Fracal pictures possess similarity and recursion, and fractal dimension naturally becomes a effective measure of similarity. The paper mainly discusses the complexity and applicability of the effective fractal dimension algorithms interested in these years, including blanket method, boxing method, fraction Brown random model method and other methods improved.