基于逆向递推的快速Otsu阈值分割算法

Fast Otsu threshold segmentation algorithm based on reverse recursion

针对Otsu法图像分割存在计算冗余问题,本文提出了一种改进Otsu阈值分割算法。首先,将图像的平均灰度值设定为逆向递推的初始阈值;其次,通过初始阈值逆向递推并推导出逆向递推公式;最后,通过递推的Otsu法在缩小灰度级范围内用背景方差替代Otsu方法的背景均值来选取最佳分割阈值。实验仿真结果表明,改进的Otsu算法缩小了计算范围,减少了运算量,提高了算法的运行速度,能满足实时性要求。

There are computational redundancy problems for Otsu method image segmentation,a method which is used for improving Otsu threshold segmentation is proposed in the study. Firstly,the average gray value of the image is set to the initial recursive threshold. Secondly,the reverse recursion formula is deduced by the initial threshold. Finally,the optimal segmentation threshold is chosen by using the background variance instead of the background mean of the Otsu method in the reduced gray level range by recursive Otsu method. The experimental results show that the improved Otsu algorithm not only narrows the computational range and spends less computation and time,but also improves the running speed of the algorithm,which satisfies the real-time requirement.