基于进化规划的最大类间方差的图像分割算法

图像分割是计算机视觉中一个关键的技术.在分析了最大类间方差算法(Ostu算法)求阈值进行图像分割以及进化规划原理的基础上,提出了一种自动阈值选取的图像分割的算法.该算法以灰度图像的灰度最大类间方差作为适应度值,将图像分割问题看作为一个全局数值优化问题,利用进化规划全局寻优和快速计算的特点,搜索一个最佳阈值用于图像分割.最后基于实例对其与传统Ostu算法进行性能比较和效果验证.结果表明,本方法在图像分割过程中具有速度快,效果好的特点.

IMAGE SEGMENTATION OF THE OSTU METHOD BASE ON EP ALGORITHM

Image segmentation is an important technique. The paper presents a new automatic threshold image segmentation algorithm after studying the Ostu threshold method and Evolution Programming algorithm. The algorithm regards the image segmentation as an optimization program that uses the maximum variance among classes as the fitness. Using the character of the Evolution Programming, the algorithm searches an optimal threshold for image segmentation. Finally, the paper compares the algorithm developed in the paper with the traditional Ostu threshold method about their segmentation qualities and computational speed. Then, it is concluded that the algorithm is not only of higher segmentation quality but also of quicker computational speed.