扩展的Otsu最优阈值图像分割的实现方法

Extended optimal Otsu thresholding method of image segmention

基于付忠良等人提出的扩展的Otsu最优阈值图像分割方法,提出了遗传算法的解决方案,并给出了遗传算法中基本参数的设定.基于图像的像素方差信息,利用遗传算法全局搜索图像的单阈值和双阈值,这样不但缩短了计算时间,而且具有遗传算法鲁棒性和自适应的特点,比传统的Otsu方法有明显的优点.

Genetic Algorithm (GA) is proposed to solve the image segmentation problem raised by Zhongliang Fu. The basic parameters of GA are defined for the image segmentation. Based on the information of average and square error of image pixels, GA is used to search globally for single threshold and double thresholds. It not only reduces the running time, but also has the robustness and self-adaptability of GA, and gains an advantage over the method of traditional Otsu.