Research On Moving Target Detection And Shadow Elimination Method Based On ViBe

Intelligent monitoring system not only provides an essential security protection for the society, but also plays a more and more important role in areas such as transportation. The popularity and optimization of intelligent monitoring system depend on its core technology, that is computer vision processing technology. Usually the first step to realize intelligent monitoring system is to detect moving targets of monitoring video, So the moving target detection technology plays a fundamental role in video monitoring system.The common methods of foreground detection are based on the background subtraction and temporal differencing technologies. Although these target detection technologies can better detect moving targets, will also encounter a series of problems, such as how to adapt to the change of the dynamic background, how to eliminate the ghost area,how to remove the shadow of moving target, ect. These are the problems of many target detection methods need to be solved. This paper mainly studies the target detection algorithm based on ViBe, and gives methods to improve the above problems. Main contents are as follow:First,this thesis simply introduces the most commonly classic foreground detection algorithm, there are three main categories of algorithms:background subtraction, frame difference and optical flow method. The thesis emphasis on the fundamentals of target detection algorithm and analyses their advantages and disadvantages respectively. By studying the three kinds of commonly target detection algorithm and regarding it as the theoretical basis. It lays a foundation for the study of moving target detection algorithm of this thesis.Then,this thesis focus on studying and researching a relative new background modeling method ViBe algorithm.Then makes a deep research and exploration of the ViBe foreground detection algorithm, also analyses several urgently need to be improved of the ViBe algorithm, such as:it easy appear ghost area; vulnerable affected by illumination change, leaves disturbance and other dynamic background; unable to suppress the shadow of moving target.Next, in view of these disadvantages of ViBe algorithm, this thesis puts forward the corresponding improvement ideas and methods. Aiming at the ghost area problem, a improved ViBe background model initialization method is came up with, accurate background image is obtained by using the thought of three frame difference method, combined with ViBe algorithm to detect moving targers can quickly eliminate ghost area; Aiming at the effects of dynamic background of foreground detection,an adaptive thershold calculative method is proposed and making match threshold R can adjust itself according to the dynamic change of background to reduce the influence for the foreground detection algorithm. Finally, using mathematical morphology method to further improve the foreground detection result. Compared with the traditional ViBe algorithm and common target detection methods, experimental results show that improved algorithm in moving target detection is more accurate.Again, researching the method of eliminate shadows of moving target, introduces several kinds of shadow elimination algorithm based on color space, studies the method of shadow elimination based on RGB color space and the color space transformation method. Then this thesis proposed a new shadow elimination method, First of all, detect preliminary shadows of foreground pixels in the RGB color space, get the set of candidate shadow pixels; Then further confirm which of them are shadow pixels, using two shadow discrimination methods:One is according to the color invariant features of shadow in C1C2C3 space, the other one is shadow discrimination method in HSV color space, then eliminate the shadow pixels which the two methods are judged, finally we achieved accurate foreground objects.