档案图像空间相关性分析  

Analysis on the Spatial Correlation of Document Image

作　　者：杨有[1] 尚晋[2] 郑琼 

机构地区：[1]重庆师范大学计算机与信息科学学院,重庆400047 [2]重庆航天职业技术学院,重庆400021 [3]中国电力工程顾问有限公司,北京100120

出　　处：《重庆师范大学学报（自然科学版）》2010年第4期65-68,共4页Journal of Chongqing Normal University:Natural Science

基　　金：重庆师范大学博士基金(No.10XLB006);云南省2009年社会发展科技计划(No.2009ZC128M)

摘　　要：对档案图像的空间相关性进行分析,有助于对档案图像的进一步处理和分析。通过使用相关系数和灰度共生矩阵这两个相关性度量准则,分别获得档案图像在像素级、像素对级和像素块级的相关性数据,在此基础上,通过对相关性数据的分布状态进行描述和解释,指出了档案图像的空间相关性特点。实验结果表明自然图像和档案图像具有明显不一样的空间相关性,具体表现在:自然图像中大量的纹理导致了自然图像的行相关性高于档案图像;档案图像结构化层次的版面导致了档案图像的列相关性分布比自然图像更加杂乱;档案图像的纹理粗糙性,导致了档案图像灰度共生矩阵的分布比自然图像更加集中。Studying on the spatial correlation of document image was benefit to the further image processing and image analyzing. Using the two measurement criteria of correlation coefficient and the gray level cooccurence matrix （GLCM）, the image correlation data of pixel to pixel, pixel pair to pixel pair and pixel block to pixel block were obtained. Based on the description and interpretation of these data, the spatial correlation features of document image were pointed out. The experiments and results showed that the spatial correlations of the document image and the natural image were different. First, because of the natural image＇ s texture, the average correlation coefficient amplitude between adjacent rows of the natural image was bigger than the document＇ s. Second, because of the document hierarchical structure layout, the correlation coefficient distribution between adjacent columns of the document image was more clutter than the natural image. And finally, because of the less image textures in document, the GLCM of document image distributed centrally.

关 键 词：档案图像 空间相关性 相关系数 灰度共生矩阵 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]