Parking Space And Obstacle Detection Research Based On Checkerboard-Grid-Laser Vision Enhancement

The emergence of automatic parking technology is a bright spot to solve the problem of parking difficulties in crowded urban parking lots.However,the accuracy of parking space identification is always a "mountain",which hinders the further promotion of parking system.The automated parking technology that uses ultrasonic radar has the problem of low identification accuracy,and this scheme is highly dependent on vehicles which parked near the parking spaces.In the vision sensor scheme,the obstacles in the parking space are seldom considered,but the size of the obstacles often affects the effectiveness of recognizing parking space.Especially,when the color of the obstacles is similar to the ground,it is difficult to recognize the obstacle.Against this background,this paper proposes a method to identify parking spaces and obstacles based on visual sensors and laser devices recognition methods by installing a laser transmitter on the car.This laser transmitter can produce a checkerboard-shaped laser grid(mesh)which varies with the condition encountered on the ground,which is then captured by the camera and then taken as the region of interest for the necessary image processing.The common obstacles in parking scenes include vehicles,stones,pits,walls,parking locks and so on and the shape of the laser grid will change significantly.Parking scene are captured by fish-eye cameras,and the vision processor completes the image processing work separately.After that the data of parking space and obstacles are sent to the parking controller through the communication serial port.This paper mainly studies the realization of laser-grid of parking space,the technology and methods of parking place and obstacle recognition based on laser device,and the architecture design of automatic parking system based on laser device.In addition,the effectiveness of the scheme is verified by simulation experiments.Firstly,the principle of the system is introduced.The two main functions of the laser grid are parking space division and size measurement.In order to realize these two functions,the calibration and capture of checkerboard-shaped laser grid(mesh)are studied.The image is captured by fish-eye cameras,so the camera model and camera calibration method are studied.Secondly,the image processing scheme based on laser grid is studied.The image segmentation and contour extraction brought by laser grid simplify the image processing.The main algorithms include image distortion correction,bird's-eye view transformation,image preprocessing,image segmentation and feature extraction.Then,the parking scene in real life is analyzed,and the types of parking spaces and obstacles are sorted out.The methods of parking space and obstacle identification are studied,and the obstacle identification,parking line identification and no parking line identification are analyzed according to the characteristics of the visual scheme based on laser device proposed in this paper.Finally,the architecture of the automatic parking system is designed,and the image processing strategy of the vision processor is proposed.The parking spot recognition and obstacle recognition experiments are carried out by simulating the real parking scene,and the different types of obstacle recognition experiments are analyzed.The experimental results show that this method can effectively identify obstacles and parking spaces as well as their size even when the obstacles and the background have a similar color compared to when only using sensors or cameras alone.