基于灰度图像的阈值分割改进方法.docx

基于灰度图像的阈值分割改进方法 Based on gray image threshold segmentation method improvement 专业班级 : 学生姓名 : 指导教师 : 系 别: 2012 年 6 月 通常人们只对图像的某个部位感兴趣，为了能够把感兴趣的部分提取出来，就得 对 图像进行分割。 图像分割就是把图像分成一些具有不同特征而有意义的区域， 以便 进一步 的图像分析和理解。 图像增强就是突出人们感兴趣有用的部分， 或者是改善图 像的质量， 使它尽可能的逼近原图像。 本论文分析了传统的灰度阈值图像分割， 即双 峰法、 迭代法和 最大类间方差法在细节部分分割上的缺点， 然后， 结合图像增强中的 微分梯度， 对原有图 像的细节进行锐化增强， 然后在使用这三种方法进行分割， 得到 的分割结果和传统的分割 方法得到的结果进行比较，该方法确实达到了改善分割后图 像细节的效果。 该方法在 matlab2008 环境下进行了实现，实验结果表明，与传统的阈值分割方 法相 比， 本文的方法不仅克服了传统阈值分割方法的不足， 而且还对复杂灰度图像的 细节部分 具有较好的分割效果，为图像分割方法的改进提供了技术支持。 关键词： 图像分割；图像增强；阈值；梯度 ABSTRACT Usually people just to certain parts of the image is interested in, in order to be able to put the interested in part of the extracted, have to the image segmentation ? Image segmentation is the image into some different characteristics with meaningful area, in order to further the image analysis and understanding? Image enhancement is prominent people interested in part of the useful, or improve the quality of the image is, make it as far as possible the original image approximation. This paper analyzes the traditional gray image segmentation threshold value, namely the petronas twin method, and the most categories iteration method in details the variance between segmentation defects, and then, combined with the image enhancement of differential gradient for the original image details to sharpen enhancement, and then use this three methods segmentation, get segmentation results and the traditional method of get results were compared, the algorithm does to improve segmentation image details after the effect. This algorithm in matlab2008 environment the realization, the experimental results show that, with the traditional threshold segmentation method than the, this algorithm can not only overcome traditional threshold segmentation method is in