Research On Detection Of Moving Object In Real Time Scene

In recent years, with the development of computer technology and image processing technology, video surveillance has infiltrated into education, government, entertainment, health, sports, security and defense areas widely. Moving objects detection is the most important issue of video surveillance, the technology is to extract moving objects from a sequence of the video image, and its goal is in order to ensuring the effect of the object classification, and tracing processing. So the results and speed of motion detection will directly affect the entire video surveillance system.In real-time environment, due to lighting changes, background interference, moving objects suddenly long stay, moving target detection results often bias. With this problem, the main research works in this article as follows:In image pre-processing, the video image is subject to kinds of noise interface in the image collection, transmission and recording process. The image must be pre-processed in order to reduce the impact of noise. The traditional methods is ideal when they only deal with single noise, but effects are not good when they deal with mixed noise. Therefore, in the article, the algorithm is proposed by adaptive weighted mean filtering. The result proves that the algorithms to be efficient. It is easy to cut out mixed noise, and at the same time, it maintains the edge of image.In moving object detection algorithm, the algorithms benefits and trouble are summarized by simulation experiments. The symmetric difference algorithm can not completely extract the moving object edge, and object internal information is also lost more serious. So this paper is proposed combined fast edge detection with the symmetry difference algorithm to improve the algorithm. The background model updates is lost, the background image is not satisfactory when the scene appears in things such as light mutation, moving objects longer stay. Those questions are solved by improving Surendra algorithm.Through the comprehensive comparison of the advantages and disadvantages of those algorithms, an algorithm has been designed combined with the two improved detection algorithms, in order to obtain more accurate and complete moving target._Simulation results show that the algorithm quickly and accurately detect moving objects and extract more information of the object. It provides accurate and reliable object information to follow-up objects tracking.