古籍书影中钤印提取技术的探讨  

DISCUSSION ON THE EXTRACTION OF SEAL PRINTS IN ANCIENT DOCUMENT IMAGES

作　　者：葛怀东[1] 尚弘 

机构地区：[1]金陵科技学院人文学院,江苏南京210038 [2]富士通研究开发中心有限公司,北京100025

出　　处：《计算机应用与软件》2017年第5期189-194,共6页Computer Applications and Software

基　　金：江苏省教育厅2014年度高校哲学社会科学研究基金指导项目(2014SJD214)

摘　　要：藏书钤印是古籍在流传过程中所钤盖的收藏印记,具有深蕴的文化内涵和艺术鉴赏价值,因此一直是古籍研究方面的重点。介绍一种基于HSV颜色空间的古籍钤印提取算法,通过该算法自动从古籍数字影像中提取印章图像。该算法包含了自适应设计,可用于不同质量的古籍文档图像的印章提取工作。同时,该方法通过颜色空间的投影分析可以有效去除正文对印章像素点所造成的干扰,从而生成清晰的印章图像。通过对200多张籍影像中的620枚印章的提取实验表明,该印章提取算法具有较好的提取效果,印章图像的召回率达到93.4%,准确率达到83.4%。Seal prints is the collection mark of ancient documents in the process of spreading,with deep cultural connotation and appreciation of the value of art,have been the focus of ancient documents research. This paper introduces an HSV color space based algorithm for ancient seal extraction. The algorithm can automatically extract the seal image from the ancient digital images,and can be used for different quality of ancient documents image of the seal extraction work. At the same time,the method can effectively remove the interference caused by the text to the seal pixel by projection analysis of the color space,and thus produce a clear seal image. The extraction of 620 stamps from more than 200 images shows that the seal extraction algorithm has a good extraction effect,the recall rate of stamp images is93. 4%,and the accuracy rate is 83. 4%.

关 键 词：古籍 藏书钤印 颜色空间 图像提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]