基于互信息最佳阈值迭代的图像分割方法研究  

A Method of Image Segmentation Based on Mutual Information and Threshold Iteration

作　　者：陈君[1] 朱红[2] 

机构地区：[1]黄石理工学院,黄石435003 [2]武汉理工大学机电工程学院,武汉430070

出　　处：《武汉理工大学学报（交通科学与工程版）》2011年第3期641-644,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：湖北省教育厅科学研究计划项目资助(批准号:D20093001)

摘　　要：阈值分割是一种重要的图像分割方法,是图像检测与识别的重要预处理步骤之一.依据分割图像与原图像之间的内在联系,提出一种互信息和阈值迭代相结合的番茄图像自动优化分割方法(MI-OPT).分割实验表明:对成熟期果实与背景的颜色差异不大或表面色彩不一致的番茄图像,MI-OPT分割方法效果较好.Threshold Segmentation is a kind of important image segmentation method and one of the important preconditioning steps of image detection and recognition,and it has very broad application during the research scopes of the computer vision.According to the internal relation between segment image and original image,a tomato image automatic optimization segmentation method（MI-OPT） which mutual information associate with optimum threshold iteration was presented.Simulation results show that this method has a better image segmentation effect on the tomato images of mature period and little background color difference or different color.

关 键 词：图像分割 互信息 阈值优化 番茄图像 

分 类 号：TP242.62[自动化与计算机技术—检测技术与自动化装置]