基于时空域多特征融合的视频拷贝检测算法  

A Video Copy Detection Algorithm Based on Temporal Spatial Domain Multiple Features

作　　者：侯艳艳[1] 李庆华[2] 王秀贞[1] 张宇 Hou Yanyan Li Qinghua Wang Xiuzhen Zhang Yu(College of Information Science and Engineering, Zaozhuang University, Zaozhuang 277160, China College of Electrical Information and Control Engineering, Qilu University of Technology, Jinan 250353, China China Mobile Group Shandong Company Limited Zaozhuang Branch, Zaozhuang 277160, China)

机构地区：[1]枣庄学院信息科学与工程学院,山东枣庄277160 [2]齐鲁工业大学电气工程与自动化学院,山东济南250353 [3]中国移动通信集团山东有限公司枣庄分公司,山东枣庄277160

出　　处：《宁夏大学学报（自然科学版）》2017年第1期33-37,共5页Journal of Ningxia University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(41204025)

摘　　要：针对视频拷贝检测中检测准确率低和误码率高的问题,提出结合时空域特征的多特征视频拷贝检测算法.该算法结合视频帧最大似然比和SURF描述子构成视频特征,较好地对抗各种视频变换,采用基于Kd树的特征匹配算法提高视频检测效率.实验表明该算法能有效提高拷贝检测的速度和准确度.To solve the problem of low detection accuracy and high bit error rate in video copy detection,a video copy detection algorithm based on spatiotemporal features is proposed in this paper.The algorithm combines video frame maximum likelihood ratio and SURF descriptors to constitute video features to deal with a variety of video transformations,feature matching algorithm based on Kd tree is adopted to improved detection efficiency of video detection.Experimental results show that the proposed algorithm can improve the accuracy and speed of video copy detection.

关 键 词：最大似然比 Kd树算法 马氏距离 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]