Research On Moving Object Detection Algorithm In Intelligent Video Surveillance

The moving target detection algorithm is an important part of intelligent video monitoring,which is a key step to realize different intelligent video monitoring tasks such as target tracking,target recognition,behavior analysis.Moving target detection algorithm is widely used in security,transportation,military,artificial intelligence,industrial manufacturing,medical,video watermark and other fields.However,the detection result of moving target detection algorithm is easily influenced by video image quality,camera shake,illumination change,similarity of moving target and background,shadow,ghost,dynamic background and so on.Scholars at home and abroad have done a lot of research in this field and have received great achievement.On the basis of reading a large number of references at home and abroad,this paper studies the moving target detection algorithms and problems in intelligent video surveillance.Firstly,three classical moving object detection algorithms are studied: time difference method,background model difference method and optical flow method.The original time difference method is essentially a two-frame difference method,which is prone to the drawbacks of the smear,dummy,and holes,therefore the three-frame difference method is introduced to solve the problem.The background model difference method is one of the most popular methods in the moving target detection algorithms,Optical flow is an algorithm that can detect moving objects under a fixed camera and under a moving camera.Secondly,Visual background extractor algorithm(ViBe)for foreground detection has the disadvantages of that there is ghosting and it is difficult to eliminate for a long time,an improved visual background extraction algorithm is proposed.Being introduced in the first n video frame sequence,the OTSU algorithm can calculate an adaptive threshold for the frame difference method and we can get a more accurate foreground region.A sample image is synthesized by using the first n frame image which is being eliminated the foreground area.The model is initialized with the extended neighborhood range in the synthesized sample images,and the ViBe background model can be updated by the extended domain.Simulation results show that the effect of ghost in foreground detection can quickly be eliminated with the improved ViBe algorithm,and foreground detection is more accurate.Lastly,the local binary similarity segmenter(LOBSTER)algorithm has poor adaptability and high detection noise in the dynamic background,and an improved LOBSTER algorithm is proposed to solve the aforementioned problems.The LBSP value of each pixel is calculated at the initialization stage of the model.The gray and LBSP values of the pixel are then added to each pixel of the color background model and LBSP background models,respectively,which enhances the description of the background model.The standard deviation,which is calculated in the neighborhood of each pixel,is utilized as a measurable index of the complexity of the pixel at the pixel classification stage.Adaptively adjusting the classification threshold in the color and LBSP background models can lower the noise in the foreground according to the background complexity.The conservative update strategy is still used in the improved LOBSTER algorithm to update the LOBSTER background model at the model updating stage.Simulation results show that the improved LOBSTER algorithm performs better than the conventional ViBe model and LOBSTER algorithm in dynamic conditions with a higher accurate rate and stronger robustness in foreground detection.