Key-Frame Based Robust Video Hashing Using Isometric Feature Mapping

In order to identify videos on the Internet, a robust video hashing based on key frames and Isometric Feature Mapping (ISOMAP) is proposed in this paper. In this method, key frames extraction and hash generation are two major components. During the video hashing, key frame extraction is achieved by a uniform distribution vector and video tomography analysis, while hash generation is accomplished by ISOMAP and hash calculation algorithm. To reduce the numerousness of frames in the high-dimensional space, video shots are first detected based on video tomography and key frames are tagged by a vector. Meanwhile, the luminance coefficients of key frames are computed to represent the high dimensional feature. The ISOMAP is carried out to globally and optimally map the key frames of video to a trajectory of points in a low dimensional space. The distances between points are computed and a hash calculation method is proposed to get hash sequence, which can inherently describe the corresponding video. Experimental results show that the video hashing has good robustness and discrimination.