Research on method of feature extraction and recognition of traffic condition from video

Route guidance information has great significance for reasonably utilizing road network resources and reducing traffic congestion. This paper proposes the image preprocessing method for traffic videos, including the background construction and adaptive image segmentation. Based on the in-depth analysis of the parameters used in road condition describing, the paper suggested the features of occupation ratio and speed of traffic stream in the observing region to characterize road traffic states, and presented the feature extracting method whose feasibility was then analyzed and demonstrated. Finally, the automatic road condition recognition was achieved based on fuzzy decision, and a higher coincidence rate was obtained.