Moving vehicle detection algorithm based on motion edge extractor

Xiaowei Hu, Jiexin Wang, Xuemiao Xu, Biao Zhou 

School of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, Guangdong, China

In this letter, we propose a moving vehicle detection method based on motion edge extractor (MEE). In the course of the vehicle detection, the motion and contour information are prominent, so we combine them together to extract the moving objects with complete contours. We first modify the Gaussian Mixture Model (GMM) to estimate the background more precisely. Then an object extraction method in static image is proposed. The original image and the background image go through the object extraction method and a series of logical operations to get the moving areas. At last we apply a simple filling method to refine the result and accurately extract the vehicle areas. The experiment result shows that our algorithm is not sensitive to illumination and can detect the vehicles with similar colour to the road robustly.