Study On Technology Of Vehicular Electro-optical Panoramic Vision Sensing System

With the sharp increasing of vehicls,traffic accidents occur frequently,collision and cut rub are the most serious in the case of traffic jam and parking at low speed.The main reason is that there exists a blind area in the field of vision,which leads drivers to obtain incomplete information about the driving environment and make wrong control decisions.In order to improve the security and standardization of driving,it is necessary to eliminate the blind areas of vision and to perceive lane lines,obstacles and other information in driving environment.In this paper,a practical solution for the system is designed by studying the Vehicular Electro-optica Panoramic Vision Sensing System,four fisheye cameras are installed around the vehicle body to collect image information,the four fisheye images collected at the same time are image-processed to generate a panoramic bird's-eye view image without the blind area,and the lane lines and obstacles in the image are perceived.The main contents of the study include the following four aspects:1)The Vehicular Electro-optica Panoramic Vision Sensing System is designed.The overall block diagram of the hardware system is presented by analyzing the requirements of the Vehicular Electro-optica Panoramic Vision Sensing System.The hardware selection and the experiment of image acquisition are completed,the software algorithm flow is designed according to the system demands and image characteristics.2)Panoramic bird's eye view image is spliced.The distortion correction,projection transformation,image registration and image fusion of the four fisheye images of the same frame are carried out based on the mosaic algorithm of the panoramic bird's eye view image,and a panoramic bird's eye view image is generated.Real car experiment is conducted to verify the effectiveness of the algorithm,the experimental results show that the algorithm produces a clear and reliable panoramic bird's-eye view image,which satisfies the requirement of no blind zone in the system.3)Perception of lane line based on panoramic bird's eye view image.By analyzing the characteristics of lane lines in images,the optimized Canny edge detection algorithm and Hough line detection algorithm are used to detect lane lines,the correct detection rate was92.92%.The lane line type identification is carried out by adopting a method of pixel statistics,The perception accuracy rate was 95.52%.The lane line RGB three-dimensional pixel information is converted to two-dimensional information,according to its characteristics the lane line color recognition is carried out,the perception accuracy rate was 93.97%.4)Perception of obstacles based on panoramic bird's eye view images.The optical flow algorithm is used to solve the vehicle motion parameters for adjusting image,and the obstacle area is acquired and identified by stereoscopic parallax,the perception accuracy rate was70.45%.