Research Of Near-duplicate Video Hashing Retrieval

With the rapid growth in video-related applications and services,such as video sharing,broadcasting,advertising,and recommendations,the number of online videos has increased exponentially.In addition,the increasing number of online user-video-related activities,including uploading,downloading,editing,and sharing,exist and have generated numerous near-duplicate videos(NDVs)on the Web.These NDVs are generated through simple reformatting,acquisitions,transformations,edits,etc.The task of the NDV retrieval(NDVR)is to return the near duplicated videos with the query video.With a large number of the NDVs,the NDVR can be applied in many fields,such as copyright protection,video monitoring,video recommendation and video retrieval.For instance,when searching for videos,we often want to find novel videos.However,the top-ranked results usually yield many NDVs.In addition,NDVs are usually generated from an original video and may infringe upon the copyright of the video producers.In the context of big data,video hashing has certain theoretical and practical significance.Multi-view hashing can well support large-scale near-duplicate video retrieval,due to its desirable advantages of mutual reinforcement of multiple features,low storage cost,and fast retrieval speed.However,there are still two limitations that impede its performance.First,most existing methods only consider local structures in multiple features.They ignore the global structure that is important for near duplicate video retrieval,and cannot fully exploit the dependence and complexity of multiple features.Second,most existing works learn hashing functions bit by bit,which unfortunately increase the time complexity of hash function learning.In chapter2,we propose a global-view hashing(GVH)framework to address the above-mentioned issue;our framework harnesses the global relations among samples characterized by multiple features.In the proposed framework,multiple features of all videos are jointly used to explore a common Hamming space,where the hash functions are obtained by comprehensively utilizing the relations from not only into-view but also inter-view objects.In addition,the hash function obtained from the proposed GVH can learn multi-bit hash codes in a single iteration.Compared to existing video hashing schemes,the GVH not only globally considers the relations to obtain a more precise retrieval with short-length hash codes but also achieves multi-bit learning in a single iteration.Furthermore,in the video hashing,the local relation can be used to reflect the relationship between the paired samples,and the global structure can be used to reflect the relationship among the whole samples.Therefore,we proposed a supervised hashing scheme,termed as joint multi-view hashing(JMVH).It jointly preserves the global and local structures of multiple features while learning hash functions efficiently.Specially,JMVH considers features of video as items,based on which an underlying hamming space is learned by simultaneously preserving their local and global structures.In addition,a simple but efficient multi-bit hash function learning based on generalized eigenvalue decomposition is devised to learn multiple hashing functions within a single step.It can significantly reduce the time complexity of conventional hash function learning processes that sequentially learn multiple hashing functions bit by bit.The proposed two method GVH and JMVH are evaluated on two public databases.The experimental results demonstrate the superior performance of GVH and JMVH compared with several state-of-art methods.