Due to the diversity of different wipe effects, wipe transition is considered complex and difficult to detect. This paper identifies two common characteristics of different wipes, which can be described by two principles – independence and completeness. By exploiting the two principles, we developed an effective wipe detector. Since no pre-determined values would be appropriate for different videos, we propose using a dynamic method to generate adaptive thresholds for wipe detection. The experimental results show that the proposed detection method can identify various wipe effects with good accuracy; overall, it outperformed other published methods.
Shan Li, M. C. Lee