Content-based Video Copy Detection

With the Continuous development of Internet technology and multimedia technology, multimedia shows explosive growth. A lot of videos is produced, circulated and edited every day. Content-based video copy detection, an important method of video organization and management and copyright protection, has become a multimedia information retrieval research focus.This paper proposed a new scale invariant interest points detector based Gabor filter and Hessian function (Hessian-Gabor Detector). This method finds the candidate interest points with the scale adapted hessian detector at first, and then get the characteristic scale of the interest points with detector based on Gabor function.The experiments show that comparing with Harris, DOG and Harris-Laplace etc, Hessian-Gabor Detector can not only reduce the computer time but also achieve better repeatability while in small scale change; HGD+GOS+FPIP can obtain close performance to other state-of-the-art descriptors like SIFT, and it can greatly improve the speed in image matching experiments. Significantly, the combination of Hessian-Gabor Detector, GOS and FPIP for video copy detection can achieve better efficiency than the other compared state-of-the-art methods.