基于混沌优化的二维熵图像分割方法

Two-Dimensional Entropy Image Segmentation Based on Chaos Optimization

对于某些灰度特征相对复杂的图像,传统的二维最大熵阈值分割法效果不佳。文章提出了基于混沌优化原理的二维熵分割法,利用类似载波的方法将混沌序列映射至双阈值的二维空间,对局部最大熵进行搜索,在找到的所有极大点中,根据目标所占图像的比例选择最佳阈值进行分割,其分割效果明显优于遗传算法所找到的全局最大值。并用此法对常用的测试图片进行了分割实验,结果证明算法具有可行性。

For the segmentation of some images with complex gray features, traditional two-dimensional maximum-entropy segmentation method can not work effectively. An approach of 2-d entropy segmentation based on chaos optimization is proposed in this paper.The method searches for all the local maximal thresholds in the way of mapping from chaos sequences to 2-d variables space which is similar to carrying waves. Then the best threshold is chosen from the points found according to the object＇ s proportion to the image. The method shows a better segmentation than genetic algrithom does. Common testing images are also experimented, the result of which approved the feasibility of the method.