Vehicle detection and segmentation in dynamic traffic image sequences with scale-rate

The scale-rate of digital signals is presented to extract local complexity feature. It is proved experimentally that the scale-rate value has direct relationship with signals´ local changing rate. Based on the scale-rate, a new method of vehicle detection and segmentation is proposed for dynamic traffic image sequences. The experimental results show the effectiveness and efficiency of the vehicle segmentation method with scale-rate, which has better performance than the frame differencing method and the optical flow method.