计算机应用与软件
計算機應用與軟件
계산궤응용여연건
COMPUTER APPLICATIONS AND SOFTWARE
2013年
12期
170-172
,共3页
图像分割%矩不变特性%最大类间差法%最大熵法
圖像分割%矩不變特性%最大類間差法%最大熵法
도상분할%구불변특성%최대류간차법%최대적법
Image segmentation%Moment invariant feature%Maximum between-class variant method%Maximum entropy method
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。
為瞭提高圖像分割算法的普適性,根據矩不變算法中圖像分割前後的亮度矩保持不變這一特性,提齣首先通過該特性來穫取閾值,然後通過該閾值來與一維最大類間差法閾值和一維最大熵閾值進行比較,併進行最終選擇的方法。通過實驗錶明,該算法能解決複雜圖像的分割問題,從而使算法的普適性得到提高。
위료제고도상분할산법적보괄성,근거구불변산법중도상분할전후적량도구보지불변저일특성,제출수선통과해특성래획취역치,연후통과해역치래여일유최대류간차법역치화일유최대적역치진행비교,병진행최종선택적방법。통과실험표명,해산법능해결복잡도상적분할문제,종이사산법적보괄성득도제고。
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 .