Research On Algorithm Of Moving Object Detection In Sequences Image

Moving target detection technology is one of the important areas in image processing and computer vision. Obtain information of moving object quickly and accurately is important in target identification, tracking, behavior understanding and describing because people are only interested in moving object or targets in detection process. Moving target detection algorithm commonly used in bright-flow method, frame differential method. background subtraction, etc. Each algorithm has both advantages and disadvantages. The algorithm of moving object detection is needed to be further studied and discussed by scholars.The paper introduces some algorithms of moving object detection based on the knowledge of the digital image preprocessing, and compared to analyze the advantages and disadvantages of each method. Then, an algorithm of combine two consecutive frame subtraction with edge difference of moving object detection is proposed. The current frame image and previous frame image are first preprocessed in the algorithm, and then frame difference operation. At the same time obtain moving region which contain moving object by calculating the threshold of the difference image and using morphological filter. Then, obtain the background image by using Surendra algorithm and the edge of the image difference by detecting the edge of current frame and background image using Sobel edge detection algorithm. For get the edge of object, the "and" operation is done between the moving region which obtain moving object and the edge of the image difference. At last the moving object is obtain through morphological processing, region filling and target identification.The simulation results show that the algorithm of the paper retains the advantages of traditional two consecutive frame subtraction detection method in stable, fast and real-time. It overcoming the disadvantages of traditional two consecutive frame subtraction method which is sensitive to the changes of the environment and light and dependents on the inter-frame interval. Especially for fast moving object relatively, the traditional two consecutive frame subtraction method can not detect the size of moving object accurately. The algorithm of the paper is practical and effective. It can remove the noise around object and the edge of background effectively. It can also achieve good results on small target detection.