Research On Fast Moving Target Detection Algorithms Used In The Highway Intelligent Monitoring System

With the coming of information era, intelligent transportation systems has played a large role in security, traffic information management, military, medical treatment and other fields. The highway intelligent highway security monitoring system as an important part of intelligent transportation, and moving target detection as the key technology of intelligent security monitoring system, the moving target detection accuracy and high or low efficiency plays a crucial role in processing the video image and relevant to the accuracy and efficiency of studying, analyzing and recognizing.According to contrasting to the existing methods of moving target detection algorithm, and through analyzing their advantages and disadvantages, this paper are proposed a kind of moving target detection algorithm applied in intelligent highway security monitoring systemIn this paper, choose three kinds of commonly used moving object detection algorithm: optical flow method, frame difference method and background subtraction method and use the OpenCV to simulate the three basic moving target detection algorithm. In the algorithm comparison process, the problem of using traditional optical flow method to detect moving object is mainly the most optical flow algorithm is relatively complex, time-consuming, poor real-time and practicality.Inter frame difference method directly subtract to the two adjacent frames, just the part which two frames relative changes will be retained and the overlap of object between two frames is not easy to be detected. Background subtraction method are particularly sensitive to changes of the scene and practically, a standard picture of the background image is always not easy to get when the background pixels in the scene changes.This paper studies a moving target detection method which combined opticalflow method and three frame difference algorithm. First, we introduce a concept of corner detection, corner is a maximum position which has a certain attribute particular point. The corner detection can find the feature points in the image, then calculate the optical flow of these feature points, thus it can simplify the complexity of algorithms and save the computing time. Due to the moving target detected is not complete, we need to add another algorithm. Because the frame difference method has a low algorithm complexity, so this paper improves the traditional frame difference method, using the three frame difference method, and then use the two difference image to carry out "and" operation. According to this way, it can improve the detection accuracy, and as a supplement to the optical flow method for moving targets were detected.Combined with optical flow method and three frame difference method of moving target detection algorithm, according to the optical flow method and frame difference method complement each other so that the system can complete extraction of the target motion and also has higher operating speed.