Automatic distinction of road surface conditions in road images at night-time using PCA and Mahalanobis weighting

The danger of causing serious traffic accidents at night-time is much higher than in the daytime. In this paper, we propose a distinction method of estimating road surface conditions by using only video information from a visible video camera. Compared to the conventional method, our innovative method provides the added benefit of the principal component analysis (PCA) of texture features and the reliability of the Mahalanobis distance. By using this method, it was possible to distinguish road surface conditions at night with high accuracy.