An Effective Vehicle Flow Counting Method

Abstract

Vehicle flow automatic counting is an important method to control effective traffic congestion. Many new methods have been proposed recently, but most are not considering the application condition in actual traffic control. For example, small vehicle spacing, sunlight changes, shadow, no movement and more are often disregarded. In this paper, a new video vehicle counting method is proposed. This method combines background modeling, frame difference, HSV dark shadow removing and Hough transformations to solve the above conditions. The proposed method can be transplanted to a high speed digital signal processor system. It is shown by numerical experiments that it gets high traffic flow detection accuracy and meets the real time requirement.