一种基于内容的文档图像检索方法

A Content-Based Algorithm for Document Image Retrieval

使用一个图像作为查询检索输入,根据该图像的版面分析特征、统计特征、纹理特征与数据库中图像的相似程度检索图像.该检索方法首先利用数学形态学对文档图像进行段落分割和行分割,作为文档图像的版面结构特征;然后根据图像的统计特征包括字符数、统计数特征、纹理特征给出文档图像抽取算法;最后给出检索算法模型.实验结果表明,本算法具有较好的查准率和查全率,在基于内容的文档图像检索中具有应用价值.

This paper studies the content-based image retrieval for document image.Given a query image,the system returns overall similar images by layout analysis and statistic feature in image database.First,segment an image into paragraphs and lines based on mathematical morphology,return the image layout analysis results;and then compute the image statistic feature include characters, statistic count feature and texture to give distil arithmetic of the document image.In the end,we describe the matching model.This algorithm is tested through trials and errors.The experiment results indicate this algorithm is good at precision and recall.This algorithm is highly valuable in document image retrieval.