基于深度学习的视频指纹快速鲁棒匹配方法  

Fast Robust Video Fingerprint Matching Method Based on Deep Learning

作　　者：李超[1] LI Chao(Department of Information Engineering,Tongling Polytechnic,Tongling Anhui 244000,China)

机构地区：[1]铜陵职业技术学院信息工程系,安徽铜陵244000

出　　处：《重庆科技学院学报（自然科学版）》2023年第2期92-96,116,共6页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

基　　金：2021年度安徽省自然科学研究重点项目“基于深度学习的视频指纹的研究”(KJ2021A1384)。

摘　　要：由于视频指纹匹配主要通过手工特征方法获取表层特征信息,导致匹配结果的调和平均值较低。为此,基于深度学习网络,设计视频指纹快速鲁棒匹配方法。针对给定视频提取梯度质心,获取待匹配视频指纹。应用卷积神经网络构建深度学习特征提取模型,得到视频指纹二值化特征,并通过平移、缩放等几何操作变换指纹细节特征。引入分层匹配理念,实现视频指纹快速鲁棒匹配。实验结果显示,此方法的匹配效果较好。Video fingerprint matching mainly obtains surface feature information through manual feature method,resulting in low harmonic average of matching results.Therefore,based on the deep learning network,a fast and robust video fingerprint matching method is designed.The gradient centroid is extracted from the given video to obtain the video fingerprint to be matched.The convolution neural network is used to construct a deep learning feature extraction model to obtain the binary features of video fingerprints,and the fingerprint details are transformed through geometric operations such as translation and scaling.The concept of hierarchical matching is introduced to realize fast and robust matching of video fingerprints.The matching effect of the designed method is good.

关 键 词：深度学习 视频指纹 分层匹配 鲁棒性 特征点 极坐标系 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]