Abstract: This paper proposes a weighting Otsu’s segmentation method based on gray probability of image from the point of view of information divergence. The weighting method based on gray probability of image is extended to obtain a generalized weighting Otsu’s segmentation method. An explanation of this new method is provided based on the fuzzy energy theory. Experimental results show that this new method is feasible, and it can help to avoid the possible inconsistent problem between image segmentation output and the real target when using traditional Otsu’s segmentation method. It provides a useful supplement to the existing Otsu’s segmentation method.

Key words: image segmentation, weighting Otsu’s segmentation method, information divergence, fuzzy energy

摘要： 从信息散度角度出发推导出一种图像灰度概率加权的Otsu分割方法，将灰度概率加权进行推广得到一般加权的Otsu分割方法，采用模糊能量理论解释其合理性。实验结果表明，加权Otsu分割方法是可行的，能避免传统Otsu分割方法提取出的目标可能与实际需要的目标不一致的问题，是对已有Otsu分割方法的有益补充。

关键词: 图像分割, 加权Otsu分割方法, 信息散度, 模糊能量

CLC Number: 

• TP391.41