Research on the Application of HOG and SVM for Wheel Recognition in Weigh-in-Motion

Detection technology based on machine vision is one of the important means of road environment perception. At present, detection technologies for vehicles and pedestrians increasingly mature, and many commercial products have been used. But machine vision in the dynamic weighing still needs to be further explored. This paper proposes a wheel recognition method based on machine vision technology used in Weigh in motion. Aiming at deal with the traditional Hough Transform is not robustness on detecting circle targets, we proposed a wheel detection method based on Histogram of Oriented Gradient (HOG) and Support Vector Machine (SVM). Firstly, HOG is used to extract the image features of the wheels. Secondly, positive and negative samples are sent to the SVM for training the classifier. Finally, the trained SVM classifier is employed for wheel detection. Experimental shows that the proposed method had higher performance compared with the traditional Hough transform, and the detection rate is over 96%, which can meet the requirements of vehicle wheel detection. In the engineering test, our method's detection rate has been improved by about 5% compared with the traditional piezoelectric wheel recognition sensor.