改进形态学相关算法以识别高相似度灰度图像  

High Similarity Gray-Scale Image Recognition with Modified Morphological Correlation

作　　者：余杨[1] 黄惟一[1] 

机构地区：[1]东南大学仪器科学与工程系,南京210096

出　　处：《光学学报》2002年第11期1362-1367,共6页Acta Optica Sinica

基　　金：江苏省重点学科经费资助课题

摘　　要：从形态学角度定义了灰度联合图像相似度 ,提出两种基于位表示法形态学相关算法的改进算法 ,通过提取位表示法的图像片边缘特征或二元化位表示法的图像片功率谱 ,以提高位表示法形态学相关算法对高相似度灰度图像的识别能力。Gray-scale joint image similarity degree is defined according to the morphology method. Two kinds of modified morphological correlation algorithm based on bit-representation decomposition are presented. The gray-scale image is decomposed into a set of binary slices in the bit-representation decomposition. In the first algorithm, the edge of each pair of slices is detected. In the second algorithm, the joint power spectrum of each pair is summed after binarized or the summation of the joint power spectrum of these pairs is binarized. Computer simulation results indicate that the modified algorithms can improve the discrimination capabilities with respect to the high similarity gray-scale images.

关 键 词：形态学相关算法 高相似度 灰度图像 联合变换相关器 图像识别 光学信息处理 位表示法形态学 计算机模拟 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] O438[自动化与计算机技术—计算机科学与技术]