Road Surface Condition Classification Based on Color and Texture Information

Abstract

Road surface condition is very important for safe driving especially in bad weather such as snow or rainy. In this paper we proposed a video camera road image status detection method. The color and texture information of the road surface is extracted from the video frame and then we build a naïve Bayesian classifier to classify the road surface image into three categories, dry, mild snow coverage, and heavy snow coverage. Meanwhile we compared the classification performance with another three popular classifiers, K-NN, Neural Network and SVM. Experimental results show that the naïve Bayesian classifier is most suitable for this classification problem.