Road Recognition Method Based On Computer Vision Research

Real-time and robust road recognition is a key technology in intelligent vehicle research, and plays an important role in vehicle autonomous navigation. As the diversity of the road and the complexity of the environment, the research on how to recognize the road quickly and accurately is a keystone on intelligent vehicle research. In recent years, some researchers made great efforts on road reorganization. They have made great progress and gained variety of methods.This paper studied a road recognition algorithm based on features. This algorithm combines the region features and the edge features to recognize the road. Firstly, the watershed algorithm based on region features is used to recognize the road. If the result satisfies our requirement, the algorithm detects the road line directly. If it does not meet our requirement, the algorithm detects the road line by edge detection algorithm. Finally, the road recognition is finished. This paper also discussed the obstacle detection based on the road detection results.The region feature algorithm is based on the watershed algorithm. This paper proposed that the algorithm updates the feature region to achieve better effect. Also, the algorithm set the updating condition.The edge feature algorithm is based on the region feature algorithm. If the division of the region feature algorithm does not meet our requirement, the edge feature algorithm is used. Edge detection is completed in the region obtained by watershed algorithm. The road recognition is finished by combining the two algorithms.Obstacle detection is another important work in this paper. Considering that obstacles are always on the road, obstacle detection only carries out in the road part.This paper established an evaluation index to evaluate the road recognition performance. The performance of the algorithm is evaluated by testing the samples in the video database. The advantages and disadvantages of the algorithm are also discussed in this paper. Experiments show that the algorithm is both effective and quick, can meet the requirement of road recognition.