The Research Of Background Difference Algorithm Based On The Model Of Pixel Filter

Vehicle overloading traffic will cause great harm to the highway, and the current efficiency of monitoring vehicles transfinite method is low. The recognition and measurement to the size of vehicle through existing scientific means has very high practical value. To solve this problem, this paper makes a basic research on the contour extraction of low speed moving objects.Based on the background difference algorithm and combined with the actual needs of overrun vehicle measurement, this paper summarize the advantages of various extract moving object contour algorithm, overcome the interference of the moving objects to the background extraction, determine the reasons for the changes of background, exclude the interference of sudden change in the background and update for long time background changes. The main research work is as follows:(1) Through the analysis of traditional background difference algorithm, Gaussian model algorithm and others improved background difference algorithm and based on the actual situation of the measurement to vehicle, this paper propose an improved scheme using pixel ratio to eliminate the interference frame. This method mainly overcomes the traditional background difference algorithm in face of the interference caused by motion object repetitive appeared in regional monitoring and the background suddenly change caused by the weather disturbed. Through these research, it improve the adaptability of detection in the real environment.(2) In order to improve the time and space efficiency of the algorithm, to overcome algorithm calculated by pixel, the concept of pixel cluster is proposed by using the experimental data. Even in face of increasingly improved image resolution can have wonderful time efficiency and the accuracy of the object contour extraction in order to meet the needs in real world.(3) Algorithm development environment based on VS2010 and OpenCV video processing library realize the background of the improved differential algorithm. Compared with experimental data of traditional background difference algorithm and others improved background difference algorithm verified the algorithm’s efficiency and accuracy for the moving object’s contour extraction.