摘要
如何实现文档图像间的快速匹配已成了人们日益关注的课题。针对文档图像的现有方法进行了研究,提出一种文档图像匹配的新方法。为进一步提高文档图像的匹配性能,结合用行列方向投影特征进行了文档图像的特征分析和提取工作,从而建立了特征相似匹配模型,在进行文档图像匹配时主要采用了平方差进行相似度度量和折半查找进行快速匹配的策略。实验表明随着二分法次数的增加,文档图像的相似度比较效率一定程度上得到很大的提高,匹配算法具有很好的抗倾斜和抗压缩效果。
How to realize the rapid matching among document images has become the people increasingly focus on the subject.Presents a new method of document image matching aiming at existing methods.In order to improve the matching performance of document images,paragraph features and row features are used to extract feature and build similar matching model in the paper.Use square error(SE)and binary search when similarity of document images are matched.The experiment shows that with the increase of dichotomy numbers,the efficiency distinctly of document images is greatly improved to some extent.Thus the algorithm is good at precision and robustness.
出处
《计算机技术与发展》
2011年第7期129-131,137,共4页
Computer Technology and Development
基金
2008年度山东省中青年科学家科研奖励基金资助项目(2008BSB38001)
关键词
相似度度量
投影直方图
平方差
折半查找
similarity measurement
projection histogram
square error
binary search