Research On The Method Of Object Exception Handling In Video Image

As the basis of image processing,moving object detection is widely used in vehicle tracking,pedestrian detection,unmanned driving and other fields,so it has become one of the research hotspots in computer vision.In the vehicle flow detection system,moving objects are easily disturbed by lighting,noise,shadow and other complex environment,which will lead to inaccurate detection results.In order to improve the detection result of moving target,ViBe algorithm which can effectively suppress noise,is used as the basis algorithm to carry out the research of abnormal situation such as ghost,incomplete target and shadow foreground in moving target detection.Firstly,this thesis analyzes the ghost in the moving foreground detected by ViBe algorithm,and it can be seen that ViBe algorithm uses the first frame to model the background.If there is an object to be detected in the first frame,the pixels of the moving object will be included when the background sample is initialized.So after the target moves,it leaves a ghost in its place.From this analysis,we can see that it is necessary to initialize the model with real background.In order to solve this problem,an improved average background method combined with ViBe algorithm is proposed.This method uses the improved average background method to get the real background and initializes the background model of ViBe algorithm with the real background.In this way,in the subsequent frame detection,there will be no ghost phenomenon.The experimental results show that the proposed algorithm can not only eliminate ghosting quickly,but also has good performance compared with other algorithms.Secondly,the effect of illumination mutation on moving object detection is studied.From the analysis of the ViBe algorithm,we can know that the target will be lost and a lot of noise will appear when the illumination changes suddenly.The reason is that ViBe algorithm uses a fixed threshold to segment the foreground.Therefore,Otsu threshold method is used to improve the fixed threshold in ViBe algorithm.Experimental results show that the improved algorithm not only solves the problem of moving target missing,but also improves the anti-jamming ability in the dynamic background.Finally,the shadow foreground of ViBe algorithm is studied,and an improved shadow removal method based on gray histogram is proposed.Since the gray histogram is based on the fixed threshold,it can not adapt to different shadow environment,so the proposed algorithm uses adaptive threshold to improve the gray histogram to eliminate shadow.But before using this method,the target image must be preprocessed.The process of preprocessing is to use the improved gray histogram better.Experimental results show that the proposed algorithm can effectively remove the shadow and has some advantages compared with other contrast algorithms.