Robust Road Region Extraction in Video Under Various Illumination and Weather Conditions

摘要: Robust road region extraction plays a crucial role in many computer vision applications, such as automated driving and traffic video analytics. Various weather illumination conditions like snow, fog, dawn, daytime, nighttime often pose serious challenges to detection. This paper presents new real-time recognition method that is able accurately extract the videos under adverse conditions. Specifically, novel global foreground modeling (GFM) first applied subtract ever-changing background frames robustly detect moving vehicles which are assumed drive region. The initial samples then obtained from subtracted model location of vehicles. integrated features extracted both grayscale RGB HSV color spaces further construct probability map based on standardized Euclidean distance between feature vectors. Finally, robust mask derived by integrating initially estimated regions located flood-fill algorithm. Experimental results using dataset real demonstrate feasibility proposed for real-time.