As vehicle driving is heavily influenced by road surface conditions, it is essential for ensuring traffic safety in winter season to properly detect the condition. Therefore, various methods such as laser radar and near-infrared detector to detect road surface conditions have been proposed. However, these methods have not been widely applied for detecting the condition of public highways because of the high deployment costs. Though on the other hand, a method using visible camera does not require expensive equipment, has been applied mainly for daytime use due to the problems of light sources. This paper proposes the method using videos acquired through an existing CCTV camera for road monitoring in order to detect road surface conditions in winter season both during daytime and nighttime. Texture features based on an RGB histogram and a support vector machine are used in this method for detecting the condition of divided into blocks. It was concluded that the method can effectively detect dry, wet and snowy conditions.
J-STAGEがリニューアルされました! https://www.jstage.jst.go.jp/browse/-char/ja/
