An improved sliding window method for shot change detection

The conventional sliding window (CSW) method has long been used in video segmentation for its adaptive thresholding technique. A hard cut is detected based on the ratio between the current feature value and its local neighborhood in the sliding window. Yet it has a relatively high rate of false alarm and missed cuts. This paper proposes an improved sliding window method which employs multi adaptive thresholds during the threestep processing of frame-by-frame discontinuity values. The proposal method uses possibility values produced from different thresholds to measure the degree of possibility of a cut’s presence. Detecting a cut based on possibility values, the method is more robust to camera/object motions. Experiment results show that our method has better detection accuracy than several other reported methods.