基于矩不变特性的阈值选择图像分割算法研究  

ON THRESHOLD SELECTION IMAGE SEGMENTATION ALGORITHM BASED ON MOMENT INVARIANT FEATURE

作　　者：李俊峰[1] 张志彦[1] 

机构地区：[1]安阳师范学院计算机教学部,河南安阳455000

出　　处：《计算机应用与软件》2013年第12期170-172,共3页Computer Applications and Software

基　　金：河南省教育厅科学技术研究重点项目(12A520001)

摘　　要：为了提高图像分割算法的普适性,根据矩不变算法中图像分割前后的亮度矩保持不变这一特性,提出首先通过该特性来获取阈值,然后通过该阈值来与一维最大类间差法阈值和一维最大熵阈值进行比较,并进行最终选择的方法。通过实验表明,该算法能解决复杂图像的分割问题,从而使算法的普适性得到提高。In order to improve the universality of the image segmentation algorithm, according to the feature of moment invariant algorithm that the intensity moment remains unchanged before and after the image segmentation, we present such a method, in it the threshold is firstly obtained through the feature, then the comparison is made with the threshold of one-dimensional maximum between-class variance method and the threshold of one-dimensional maximum entropy through the obtained threshold, and finally selection is decided. Experiments show that, the algorithm can well solve the segmentation of complex image, so that the universality of the algorithm has been improved as well.

关 键 词：图像分割 矩不变特性 最大类间差法 最大熵法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]