Abstract: Because of the high computational complexity of classical 2D Renyi entropy thresholding, a new 2D Renyi gray entropy image thresholding is proposed. This method calculates two optimal thresholds of 1D Renyi gray entropy algorithm independently instead of the optimal threshold of a 2D Renyi gray entropy algorithm. It proves that the two are equal when some condition is satisfied, and its computational complexity is reduced from O(L4) to O(L), the computational time is only one ten-thousandth of the computational time than of the classical method.

Key words: image segmentation, threshold, Renyi entropy, decomposition

摘要： 针对传统二维Renyi熵阈值法的高计算复杂性，提出一种新的基于分解的二维Renyi灰度熵阈值分割方法。该方法通过求解两个一维Renyi灰度熵阈值替代二维Renyi灰度熵的最佳阈值，理论上证明当满足一定条件时，两者等价；同时将计算复杂度由O(L4)降到O(L)，所耗时间约为传统二维Renyi熵算法的1/10 000。

关键词: 图像分割, 阈值选取, Renyi熵, 分解