Weber local descriptor from three orthogonal planes based video copy detection

This paper presents a novel video copy detection system. The kernel of the approach is based on our proposed extended local descriptor WLD to three orthogonal planes (WLD-TOP). Indeed, in the aim to extract features vector, key-frames are generated and then a perceptual hash is performed using the WLD-TOP descriptor. The proposed method is applied on three databases and evaluated against several attacks. Performances are assessed in relation to recent studies of the state of the art and the use of other descriptors. The obtained results outline the robustness of the proposed video copy detection in terms of precision, recall and accuracy.