基于灰度直方图的多阈值分割法  

A MULTI-THRESHOLD SEGMENTATION METHOD BASED ON GRAY-SCALE HISTOGRAM

作　　者：刘欣欣[1] 李雪[1] 王琼[1] 

机构地区：[1]南京理工大学计算机学院,江苏南京210094

出　　处：《计算机应用与软件》2013年第12期28-30,63,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61005005);中央高校基本科研业务费专项资金项目(NNUST2012XQTR002)

摘　　要：为了提高和增强多目标图像的分割效果,在Otsu法进行阈值分割的基础上,结合直方图的一些特有的性质,提出一种新的多阈值分割的方法。该方法利用类间方差最大原则,以二分法为基础,并将直方图上的极小值点作为多阈值分割的参考点。通过比较不同算法的分割效果,在大量的图像上进行了实验。结果表明,该算法在多目标图像分割上适应性较好。In order to improve and enhance the segmentation effect of multi-objective images, on the basis of threshold segmentation with Otsu and in combination with some particular properties of histogram, we propose a new multi-threshold segmentation method. The method uses the principle of maximum between-class variance, takes the dichotomy as the basis, and considers the minimum points on the histogram as the reference points of threshold point. By comparing the segmentation effects of different algorithms, we carry out the experiment with a large number of images. Results show that the proposed algorithm has better adaptability in multi-objective image segmentation.

关 键 词：多阈值 OTSU法 图像分割 二分法 极值点 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]