摘要： 针对基于互信息的图像分割方法对噪声干扰或光照不均匀的图像不能获得满意分割效果的不足，该文提出模糊化互信息的图像分割新方法。该方法将现有互信息进行模糊修改，得到了模糊互信息，在图像分割中给出模糊隶属度函数的定义方法，获得了模糊互信息参数的选取办法。实验结果表明了该模糊互信息图像分割方法的有效性。

关键词: 图像分割, 阈值化, 互信息, 模糊隶属度

Abstract: Considering that the image segmentation method based on mutual information can not obtain ideal segmentation quality on the image with noise or uneven illumination, the new segmentation method based on fuzzy mutual information is proposed. The mutual information is fuzzily modified into the fuzzy mutual information, then it is applied in segmenting image, and the definition method of fuzzy membership function of fuzzy mutual information is given, the method of choosing parameters of fuzzy mutual information in segmenting image is proposed. Experimental results show that the method is more feasible than that of classic mutual information.

Key words: image segmentation, thresholding, mutual information, fuzzy membership

中图分类号: 

• TP18