Abstract: A background-color-invariable-based improved twin-histogram algorithm is presented to detect shot boundary in this paper. Firstly, a fourth-orders statistic hypothesis testing in inter-frame difference is used to separate background from a general video sequence. Secondly, (L^*a^*b^*) color model is used to calculate the hue in the background areas. Lastly, the improved twin-histogram algorithm is used to detect shot boundary. This method needs fewer calculations and has a higher precision ratio. The experimental results show that the method is more accurate and effective in shot detection than other algorithms.

Key words: shot boundary detection, hue histogram, twin-histogram algorithm, cut model and flashlight model detector