Shot-based temporally respective frame generation algorithm for video hashing

In this paper, we proposed shot-based temporally respective frame (TRF) generation algorithm for video hashing. The edge change ratio (ECR) is adopted to measure the dissimilarity between different video frames. The maximum value of ECR inside a temporal sliding window is compared with a given threshold to determine the shot boundary. We weight the frames in the same shot to generate the temporally respective frame (TRF). In the experiment, the proposed TRF algorithm is applied to three existing video hash algorithms to replace the traditional temporally informative respectively image (TIRI). The experiment results demonstrate that video hash extraction from the temporally respective frame (TRF) based on shot can enhance the recall rate and preserve the high precision rate for video copy detection.