基于多阈值算法融合的图像分割

Image segmentation based on multi-threshold fusion

针对一种阈值分割算法很难对不同类型的图像进行有效地分割的问题,提出一种多阀值算法融合的方法.该方法通过不同的算法得到一组阈值,由迭代的方式将它们有机地组合,并定义一个回报函数,来选取最优阈值.实验结果表明,该方法针对不同类型的图像能够选择最优的阈值进行有效的分割,消除了使用者必须具有先验知识的障碍,省去了人工选择分割方法的麻烦.

Aiming at the problem that one kind of threshold segmentation algorithms was difficult for different images for effective segmentation,a multi-threshold algorithm fusion method based on introducing several common threshold segmentation algorithms was presented. This method got a set of thresholds by different algorithms and got a combination of them by organic iterative manner,and defined a reward function to select the optimal threshold. The experiment results showed that this method could be used for different types of image to get optimal threshold value for effective segmentation,eliminate the barriers which the user must have prior knowledge and eliminate the trouble of manually selecting segmentation method.