中值邻域二维最小交叉Tsallis熵的快速图像分割  

Fast Image Segmentation Based on 2-D Minimum Cross Tsallis Entropy with Median Value Neighborhood

作　　者：张新明[1] 陶雪丽[1] 郑延斌[1] 张慧云[1] 李双[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,河南新乡453007

出　　处：《电光与控制》2011年第5期28-33,共6页Electronics Optics & Control

基　　金：河南省重点科技攻关项目资助(092102210017;102102210180);河南省教育厅科技攻关项目资助(2008B520021)

摘　　要：鉴于邻域窗口影响二维阈值法的分割结果,提出了一种基于中值邻域二维最小交叉Tsallis熵的快速图像分割方法。首先利用中值滤波法构建中值邻域二维直方图;然后将最小交叉Tsallis熵运用在这种直方图上构建中值邻域二维最小交叉Tsallis熵分割法,由于中值滤波后的图像优于均值滤波后的图像,此法能获得更理想的阈值;最后将递推法与定义的数组运算相结合导出快速算法搜索最佳阈值向量,并用此阈值向量对原图像和中值邻域图像进行分割,得到更好的分割结果。实验结果表明:相对于当前均值邻域二维最小交叉Tsallis熵阈值法,该方法不仅分割效果更好,抗噪性更强,而且速度更快。A fast image segmentation method based on 2-D minimum cross Tsallis entropy with median value neighborhood was proposed in this paper considering that the segmentation result is influenced by the neighborhood model.Firstly,a two-dimensional histogram with median value neighborhood was established by utilizing the advantage of median filters.Then,the minimum cross Tsallis entropy was used on the 2-D histogram to get a more ideal threshold.Finally,the defined array operations and the recursive approach were combined together to get a fast recurring algorithm for searching the best threshold vector,which was used to segment the original and the median-filtered images for obtaining better segmentation effects.Experimental results showed that the proposed method can achieve better segmentation results and better anti-noise performance with less computation time than the current thresholding method based on 2-D minimum cross Tsallis entropy with average value neighborhood.

关 键 词：图像分割 二维最小交叉熵 TSALLIS熵 中值邻域 递推算法 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] TP391.4[自动化与计算机技术—计算机应用技术]