The motion detection method apply to video-surveillance embedded system

In embedded video surveillance system, many moving target detection methods are used. But for some complex environmental, they have some problems in detection accuracy and speed of moving target detection. In this paper, a target detection method based on improved adaptive threshold hybrid difference is proposed, then for wiping off the noise, the method of morphological filtering is adopted to gain exact moving objects. Finally, according to the requirement, the area method is used to choose the final images to avoid false alarms. Experimental results show that the proposed method not only gets moving objects completely, but also adapts to the complex environmental change, at the same time, the method is rapid, efficient, and is applied to embedded system well.