With the exponential growth of multimedia data, people are overwhelmed with massive amount of online videos, of which Near-Duplicate Videos (NDVs) occupy a large portion. In this paper, we present a novel framework for NDV retrieval, which explores the parallel power of two promising techniques: Graphics Processing Unit (GPU) and MapReduce. With the power of the proposed framework, various key algorithms in the field of computer vision, such as K-Means clustering, bag of features, inverted file index with hamming embedding and weak geometric consistency, are applied to NDV retrieval. Experimental results on the benchmark CC WEB VIDEO NDV dataset demonstrate that the proposed framework can significantly speed up processing huge amounts of video repositories.

No abstract

In recent years, most advanced image retrieval algorithms are built upon local features, and various up-to-date match kernels are developed to boost image retrieval performances. However, most of these image retrieval algorithms need to face up two challenging issues: (1) the locality property of local features as well as quantization noise and (2) the phenomenon of burstiness, which significantly affect image retrieval performances. In this paper, two novel techniques including Twin Feature (TF) and Similarity Maximal Matching (SMM) are proposed for image retrieval performance improvement, which can be employed with non-aggregated kernel models, for example, the Selective Match Kernel (SMK). The proposed TF employs extra information from neighboring image patches to refine visual matching. As far as SMM is concerned, it tries to control burstiness by dynamically searching the match-pair combinations to maximize the global similarity score and thus removes multiple matches. Experimental results on two benchmark image datasets including Oxford5k and Paris6k demonstrate that the new techniques SMK tf (SMK with TF) and SMKsmm (SMK with SMM) can greatly enhance image retrieval accuracy performances as compared to SMK, and their combination, i.e., SMK tf+smm , is able to achieve better image retrieval accuracies than a number of state-of-the-art approaches.