Research On Moving Target Detection Algorithm Based On ViBe And Multi-feature Fusion

The main purpose of moving object detection technology is to extract moving objects and provide accurate and effective information for subsequent target recognition and tracking.It is widely used in medical,military and transportation fields.However,in complex scenes(such as background changes)the accuracy is greatly reduced,so it is important to study an efficient and accurate algorithm for moving target detection.In order to extract moving objects accurately,this thesis takes Vi Be algorithm in moving object detection of video images as the main research object,and the work is summarized as follows:Firstly,an adaptive Vi Be algorithm is designed to solve the problem that the Vi Be algorithm is susceptible to background changes such as illumination,false detection of slow moving objects as backgrounds,and ghosts when there is a moving object in the first frame.This method introduces standard deviation rate to change the matching radius adaptively and updates the rate automatically by using the inter-frame average speed measurement.The optimum threshold for the integrity of the image to be measured is obtained by iteration method for quadratic discrimination to suppress ghosts.The experimental results show that the algorithm can better overcome the impact of dynamic background,improve the accuracy and suppress ghost phenomenon.Secondly,to solve the problem that Vi Be algorithm is difficult to distinguish shadows from moving objects,this thesis presents a shadow detection algorithm based on multi-feature fusion.This algorithm uses the foreground target obtained by adaptive Vi Be algorithm to detect HSV,LBP and Gabor filtered shadows respectively,and obtains three shadow detection results.The three results are used to get shadowed areas using the fusion strategy,and the post-processing operation is used to remove noise and further correct the results.The experimental results show that the algorithm in this thesis has obvious advantages over other algorithms in detecting shadow based on the analysis of detection effect and objective evaluation index.Thirdly,in order to solve the incomplete extraction problem of Vi Be algorithm in complex scenes,this thesis combines the adaptive Vi Be algorithm with shadow detection algorithm to design a new moving target detection algorithm,M-Vi Be algorithm.The experimental results show that the algorithm can overcome the impact of dynamic background changes and detect shadow areas,and extract targets accurately.