A Robust Weber’s Local Descriptor for Copy Control

Simple and robust local powerful descriptor called Weber Local Descriptor (WLD), which is put forward for detecting duplicated videos. Matching based on orthogonal moments has high computational complexity, especially for high dimension. Thus, to reduce the time complexity video is segmented into frame by twin threshold segmentation and key frames are extracted from each segment. Then, WLD descriptor is incorporated for feature extraction from each frame. Specifically, WLD consist of two components one is excitation and other is orientation. We use both these components to construct WLD histogram feature.Finally matching takes place between the features extracted from the reference video and the target video to check whether it is a duplicated or original video. The proposed method is more powerful to different noises, photometric and geometric transformations in the frame and detect the copied videos effectively with less computational complexity.