A 3D wavelet and motion vector based method for wipe transition detection

The detection of transition between shots in video programs is a key step in video content analysis. Wipe detection is more difficult than the detections of abrupt and other gradual transitions due to its various patterns and the difficulty in discriminating a wipe from camera and object motion. In this paper, a method for wipe detection is proposed based on three-dimensional wavelet transform and motion vector. Global motion compensation is presented with Gaussian weighted Hausdorff distance to restrain the effects of camera and object motions. This method can not only detect wipe transitions and their boundaries but also determine the types of wipes and restrain varies motion effects effectively. Furthermore, this method does not need any threshold in the wipe detection, which reduces the influence of people interaction. The experimental results on some video programs from the CCTV show the effectiveness and robustness of the algorithm.