Abstract: A fast shot boundary detection method based on visual kernel region histogram is proposed in this paper.The method consists of pre-processing,cut detection and gradual transition detection.In the pre-processing round,the even blocking of kernel region based on human visual features is presented and the adaptive local thre-shold is adopted to filter out non-boundary segments.Cut detection is carried out by the use of histogram distance.According to the color histogram change rule of gradual transition,an approach combining the histogram entropy difference with its outliers is used in gradual transition detection.The experiments indicate that the computational complexity can be reduced significantly,and the detection results can be improved on both abrupt and gradual shot boundary by our method.