Dynamic Video Background Updating Algorithm Study

In recent years, video surveillance in the production and life of the neutral is becomingmore and more important, so people pay more attention. Traffic safety, public securitydetection, safety, mall security, residential security, video surveillance throughout theproduction and all aspects of life. The moving object detection video surveillance as animportant goal to become the focus of research. The purpose of the moving object detection isto find changes in the video sequence from the region, thereby separating the moving target.The partition of the region of the changes region is the basis of latter part target identification,target tracking and behavioral of understanding. So the efficiency and the effect of targetdetection in video processing is vital. Motion video background subtraction as a simple andeffective way to get promoted in traffic monitoring, surveillance and other aspects of the bank,in behavior analysis and visual inspection is an important adjunct. Background subtractionincludes video image preprocessing, background modeling, moving object detection andbackground update these four main steps. In background subtraction, the key to moving targetdetection is getting a complete and stable background, therefore the dynamic videobackground update becomes the focus of the study.Firstly, the background difference method are detailed introduced, and analyzed theprinciple of several typical background difference method and the advantages anddisadvantages of it. To extract the moving targets in heavy traffic environment, proposes aprediction method based on analogous median filter pixel detection. This method uses adual-threshold threshold segmentation method as the foreground detection methods, and theuse of bacterial foraging optimization algorithm to optimize and improve computing speed,moving object segmentation effect increased significantly, basically meet the requirements ofreal-time monitoring. For fast traffic situation presents a Kalman filter background blockclassification based optimization algorithm updates, background updating model proposedsecondary block under three categories, get a better video background.This algorithm using Matlab and C language programming. Video capture traffic underdifferent lighting conditions at different times, after many experiments. The results show thatthe proposed motion video background updating algorithm can meet the requirements ofreal-time and accuracy, high-volume traffic environment in target detection effect is obvious,clear objectives and complete. Kalman filter algorithm based on block classification, timeefficiency and the background obtained better results than the original algorithm.