基于特征匹配的绘画作品抄袭判定  

Plagiarism Check of Painting Works Based on Image Matching

作　　者：何灵璐 魏宏安[1] 赵铁松 HE Linglu;WEI Hongan;ZHAO Tiesong(College of Physics and Information Engineering,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福州大学物理与信息工程学院,福建福州350108

出　　处：《福建师范大学学报（自然科学版）》2020年第6期36-41,共6页Journal of Fujian Normal University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61671152)。

摘　　要：当前在绘画作品领域存在的抄袭现象,通过人工进行一一判别十分困难.传统图像匹配技术主要针对严格场景的判定,而不适用于相似场景,如绘画作品在内容、风格上的抄袭.针对相似场景的匹配,提出一种基于特征匹配的绘画作品抄袭判定的算法.该算法首先使用FAST算法提取绘画作品特征点并进行匹配,进而借助Hough变换获取高层空间特性进行相似性判定,最终实现绘画作品的抄袭判定.经测试验证,该算法查全率平均可达到85.28%,查准率平均可达到85.00%.At present,there are a lot of plagiarism in the field of painting works and it is very difficult to discriminate one by one manually.Traditional image matching techniques are mainly used to judge strict scenes,but are not applicable to similar scenes,such as copying the content and style of painting works. Aiming at the matching of similar scenes,this paper proposes an algorithm for judging the plagiarism of painting works based on feature matching. This algorithm firstly uses FAST algorithm to extract the feature points of painting works and match them,and then obtains the spatial characteristics of high level with Hough transformation for similarity judgment,and finally realizes the plagiarism judgment of painting works. The test results show that the recall ratio of the algorithm can reach 85.28% and precision ratio can reach 85.00% on average.

关 键 词：绘画抄袭检测 图像匹配 计算机视觉 图像处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]