Arimoto熵是一种广义熵形式.本文首先指出了已提出的二维Arimoto熵阈值分割法的表述错误, 给出了正确的二维Arimoto熵阈值分割法; 然后提出了二维Arimoto熵直线型阈值分割法, 并给出了快速递推公式; 对Arimoto熵公式中参量的选择进行了探讨, 并基于标准图像进行了分割性能评估.大量分割实验表明, 二维Arimoto熵直线型阈值法至少与二维Arimoto熵和二维Renyi熵直线型阈值法分割效果相当; 在图像边缘和噪音信息丰富的情况下, 二维Arimoto熵直线型阈值法的分割效果优于二维Arimoto熵和二维Renyi熵直线型阈值法, 是一种有效的图像阈值方法.

Arimoto entropy is a general form of entropy. Firstly, a representation error on the two-dimensional Arimoto entropy is pointed out, and a correct two-dimensional Arimoto entropy thresholding method is given; a two-dimensional Arimoto entropy linear-type thresholding method and its fast recursive formula are proposed; Arimoto entropy formula parameter selection and the segmentation performance assessment according to the ground truth images are discussed. A large number of segmentation experiment results show that the two-dimensional Arimoto entropy linear-type thresholding method has at least a similar effect with the two-dimensional Arimoto entropy & the two-dimensional Renyi entropy linear-type thresholding; in the cases of the more image edge and noise information, the two-dimensional Arimoto entropy linear-type method is better than the two-dimensional Arimoto entropy & the two-dimensional Renyi entropy linear-type method, is a effective thresholding method.