Traffic light recognition based on spectral residual model and multi-feature fusion

Abstract

Traffic light detection and recognition play an important role in Advanced Driver Assistance Systems and driverless cars. This paper proposed a new method based on spectral residual model and multi-feature fusion to solve the problem of traffic light recognition. First, the image acquired by the camera is converted to the LAB and HSV color space, and the A-channel and S-channel are used to obtain a saliency map through the spectral residual model. Secondly, using prior information of traffic light establish a task model for determining candidate area. Then extract the HOG features, LBP features, and RGB color features of the candidate areas, and after multi-feature fusion, the traffic light status is recognized by the SVM (Support vector machine) classifier. The experimental results show that the recognition rate of the algorithm reaches 96%, which can provide stable and accurate traffic light status information for driving vehicles.