一种基于均值漂移和遗传算法的图像分割算法  

An Image Segmentation Algorithm Based on Mean Shift and Genetic Algorithm

作　　者：王建[1] 张建伟[1] 陈仲恒[1] 

机构地区：[1]四川大学国家空管自动化系统技术重点实验室,四川成都610065

出　　处：《计算机技术与发展》2014年第4期33-37,共5页Computer Technology and Development

基　　金：国家"863"高技术发展计划项目(2012AA011804)

摘　　要：均值漂移算法在图像分割中有着重要的应用,考虑到图像的噪声和边缘保持等因素,提出了一种基于均值漂移的图像平滑和利用遗传算法取得直方图最大熵的图像分割算法。首先利用均值漂移算法对图像进行平滑处理,然后对平滑后的图像通过遗传算法获得在直方图最大熵时的分割阈值,均值漂移平滑后的图像不仅能很好地保持图像的边缘特征,还能有效地去除噪声,而利用遗传算法可以比较快速和准确地取得分割阈值。实验得出,文中算法能够取得较好的分割效果。Mean shift algorithm has important applications in image segmentation, taking into account the image noise and other factors such as edge retention, present an image segmentation method based on mean shift to smooth image and genetic algorithm to obtain the histogram maximum entropy. Firstly smooth image with mean shift algorithm, then obtain the segmentation threshold under the condition of maximum entropy of the histogram through genetic algorithm based on smoothed image, the smoothed image after mean shift not only keeps the edge features, but also filters the noise effectively, applying the genetic algorithm to obtain segmentation threshold quickly and accurately. The experiment shows that the algorithm can get better segmentation effect.

关 键 词：均值漂移 遗传算法 图像分割 最大熵 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]