Technical Research Of Vehicle Collision Avoidance Warning System Based On Monocular Vision

This paper studies vehicle detection based on monocular vision and discribes different methods for vehicle detection and vehicle tracking in the daytime and nighttime.It puts forward improved algorithms to solve actual problems acccording to the characteristics of different methods.In the daytime,Advanced Driver Assistant System is very demanding for safety,In order to improve the security of the system,this paper propose a new mechanism that setting the corresponding security level for different regions of interest and besides two detection methods are used successively in the alert region of interest.In addition,Lane line recognition is very important,In order to enhance the flexibility of traditional lane detection algorithm,this paper propose a new superpixel-based lane line detection algorithm,which is robust to the occlusion of the lane line.Finally,In order to improve the precision of the system,this paper has designed a PCANET-based vehicle classifier.In the nighttime,a new vehicle detection mechanism in mixed traffic scenario is designed to detect the front car or electric vehicle.Traditional cameras are automatically exposed,which makes the bright block wrongly detected as a taillight easily,this paper proposed a new mechanism that lower exposure of the camera,then segment out the red halo from the frame using color threshold The algorithm of Dempster-Shafer are used to verify the detection results of the tail lights.The experimental results show that the new algorithm reduces the vehicle detection error and improve the security of the system.In the vehicle tracking,Aiming at the difference between day and night images,a corresponding tracking algorithm is designed.In the daytime,In order to improve the accuracy of tracking,a method combining Meanshift and intelligent tracking queue is designed.In the nighttime,Aiming at the disadvantage that traditional tracking algorithm can not be applied to low brightness images,a method combining night detection model and tracking queue is designed to improve the processing speed of the algorithm and reduces the vehicle detection error.The result shows that when the algorithm works at the system of Windows 10 and 2.4 GHz CPU,the detection speed can achieve 20 frames per second and the algorithm can detect and track front vehicles accurately.