Traffic Monitoring System Moving Target Detection And Classification

Intelligent video surveillance system is related to image processing andcomputer vision field of the multiple disciplines including research subject, itbesides has important scientific meaning besides, in human-computer interactionsystem, intelligent transportation system, traffic detection, vehicle identificationand military field are widely application value and the market prospect. Targetmotion detection and classification technology is intelligent video surveillanceof the two basic core technology, they are the follow-up of various seniorprocessing, such as target tracking and behavior analysis, incident detection,video image compression and semantic indexing and high level of processingand application's cornerstone, it is also the key.of video monitoring system,automatic and intelligent and real-time applications. However, the dynamicbackground,the change of illumination,the moving shadow and the occlusionphenomenon may produce significant impact to the result of motion detection.To find out the deal with complex environment and all kinds of changes,accurate, rapid and steady detection and classification to carry out the algorithmis still moving target urgently to be solved the issue.This paper, with intelligent video surveillance, based on the video monitorto key research in motion target detection and classification issue. Currentmovement of objects in the visual analysis problem faced for the foothold, of themovement of the target recognition algorithm and classification algorithm offurther research, and according to the actual need of the system, the rate,robustness and speed, the technical indexes of achieving the requirements.Moving targets examination for the movement targets detection algorithmsare summarized, based on the background of the update Surendra backgrounddifference method and five Frame difference method, further analysis, frombasic theory to specific application in this paper. And verified by theexperiments based on the background of the update Surendra backgrounddifference method and five Frame difference method of the efficiency of thealgorithm. The target motion categories, first in feature selection choices several to thearticle effectively shape characteristic, then deep into the study based on supportvector machine classifier algorithm, based on support vector machine classifierlearning and research found that using an improved method of kernel function,can improve the classifier of real-time and classification accuracy. This paperbased on wavelet, some advantages choose the based on wavelet kernel functionof support vector machine classifier, and will it compared with the traditionalmethod, to verify its validity.