Extracting Vehicle Density From Background Estimation Using Kalman Filter

Abstract

This study aims to extract traffic density through traffic monitoring camera images. To this end, it uses Kalman filter based background estimation, which can efficiently adapt to environmental factors such as light change. The difference between the incoming image and the calculated background was subjected to the proposed filters and the vehicles in the foreground were marked. The binary image representing the background and the foreground was subjected to geometric correction to equalize the effects of the vehicles near and distant to the camera, and then, the road ratio of the vehicles was computed by proportioning the foreground to the entire road area. All these procedures were applied to the road areas manually marked beforehand. The developed method was experimented on four different points recorded by traffic surveillance cameras operated by the Traffic Control Office of Istanbul Metropolitan Municipality.