Enhancement of dissolved shot boundary detection with twin-windows amplification method

We propose an algorithm to detect shot transitions of video data in the compressed domain and then extend its effect to the spatial domain. This algorithm is an improvement of the conventional solution, where energy histograms of the DC coefficients are used to calculate the distance between consecutive frames. The DC coefficients of compressed MPEG video data are obtained either through direct extraction [intraframes (I-frames)] or by frame conversion [predicted frames (P-frames)] while those of uncompressed video data are calculated by 8×8 block pixel averaging. The key contribution of this approach is to enhance the detection result by using the ratio between two sliding windows to attenuate the low-pass-filtered frame distance, resulting in the amplification of the transitional regions. The experimental results demonstrate that this algorithm achieves high detection rates with low computational complexity.