改进的Otsu算法在粮食颗粒图像分割中的应用

Grain Kernel Image Segmentation Based on an Improved Otsu Algorithm

针对二维Otsu自适应阈值算法计算复杂度高的问题,采用二维直方图投影到对角线得到一维阈值的方法,消除二维自适应阈值算法中的冗余计算,从而提高了计算速度。为此,将该方法应用于粮食颗粒图像分割中。实验表明,该算法在计算时间上远小于原始二维Otsu算法。

This paper gives a fast algorithm for two - dimensional Otsu adaptive threshold algorithm that overcomes the disadvantage of computational complexity. By the method of projecting the two - dimensional histogram onto the diagonal, this fast algorithm gets rid of redundant computation and obtains one - dimensional threshold, so the calculation speed is greatly improved. This method is applied to grain kernal image segmentation, the experimental result shows that the computational time of the method is far less than that of the origianl two - dimensional Otsu algorithm.