基于NSCT-GLCM的印章图像特征提取  

Feature extraction of seal image based on GLCM in NSCT domain

作　　者：余彪[1] 万水龙 刘进[3] 赵爽[1] 

机构地区：[1]中国人民解放军92493部队 [2]中国人民解放军91960部队 [3]南京航空航天大学

出　　处：《微型机与应用》2014年第4期29-31,34,共4页Microcomputer & Its Applications

基　　金：文件检验鉴定公安部重点实验室开放课题(10KFKT005)

摘　　要：为了有效提取印章图像的特征,提出了一种基于NSCT域GLCM的印章图像特征提取方法。首先对图像进行NSCT分解,得到多幅子图像;然后通过GLCM提取图像的特征信息,并用特征参数描述该矩阵,用该特征参数构成的特征向量代表图像的特征信息;最后利用欧式距离对图像进行分类。针对印章图像进行了大量的实验,并与GLCM法和LBP-GLCM融合法进行了比较。结果表明,该方法识别率更高。In order to effectively extract the seal image characteristics, feature extraction method of seal image based on GLCM in NSCT domain is proposed in this paper. Firstly, seal image is decomposed by NSCT to obtain multiple subband images, and the characteristic information is extracted according to GLCM, the characteristic parameters are used to describe the matrix. Use the feature vector made of the characteristic parameters to represent the feature information of images. Finally, different types of seal images are distinguished by Euclidean distance in the feature space. Experimental results on the seal image show that compared with the GLCM and LBP-GLCM fusion method, the method proposed in this paper has a higher recognition rate.

关 键 词：印章图像 特征提取 非下采样CONTOURLET变换 灰度共生矩阵 

分 类 号：TN911.73[电子电信—通信与信息系统]