Approching Obstacle Detection Based On Fisheye Camera

With the continuous development of social economy,the number of vehicles is increasing with a huge number.Therefore,the probability of traffic accidents is increasing.How to reduce the accident rate,how to help drivers make correct judgments in emergencies,and how to design a more intelligent auxiliary driving system are the hot topics of our current research.Fisheye camera has the advantages of abundant information and low installation cost,so it plays an irreplaceable role in auxiliary driving system.Therefore,the vehicle detection system studied in this paper is based on fisheye camera.The image correction of fisheye camera,moving target detection and the integrity of obstacle area are the focus of our current research.Firstly,a feature point based moving object detection algorithm is proposed.Firstly,a feature block selection method based on gradient maxima is proposed,which enables features to be distributed as far as possible on the outer and inner edges of obstacles and texture edges,while the number of selected features is limited.The time complexity of the algorithm is greatly reduced,and then feature block classes are clustered to obtain complete obstacle regions.The experimental results show that the proposed method can detect targets in various situations,and the accuracy and robustness of the proposed method are improved compared with the current mainstream algorithms...