Design Of Blind Zone Monitoring And Visualization System For Automobile A-pillar

A-pillar is the column on each side of the front windshield of a car that connects the hood and the roof.The structural strength of the column directly affects the safety of the driver when the car confronts the strong physical shock in front.Therefore,the car A-pillar has very strict design specifications,which cannot be designed too thin.However,the A-pillar with large width will block the driver’s sight and produce a blind zone,causing serious safety risks.In this article,under the premise of retaining the certain structural strength of A-pillar,the blind zone monitoring and visualization system for automobile A-pillar will be designed.This system will assist the driver to know the scene in the A-pillar blind zone,and thus the accident risks caused by it are greatly reduced.The research contents of this article consist of A-pillar blind zone visualization system and A-pillar blind zone monitoring system.The A-pillar blind zone visualization system displays the scene from the driver’s perspective by installing a display inside the car and a camera behind the A-pillar,respectively.At the same time,employing the method of 3D reconstruction,the stereo camera installed in the front is capable of tracking the sight direction of the driver.Then,the perspective transformation algorithm is used to dynamically adjust the scene displayed on the display.After the transformation of the scene,adjustment and cutting will be performed,and the transparent effect of the A-pillar is achieved to a great extent.The A-pillar blind zone monitoring system obtains the distance to the obstacle through the ultrasonic sensor outside the car.It also gets speed and turn signal information from the car ECU.The monitoring system only turns on the visualization system when necessary to reduce visual fatigue caused by the constant light of the A-pillar display.Firstly,this article studies the method to improve the camera calibration accuracy.Based on Zhang’s calibration method,a method to quantitatively determine the pose combination of the calibration board in the calibration process is proposed.This method employs the smallest singular value of the identification Jacobian matrix,which correlates the calibration error and the checkerboard pixel positioning error,as the evaluation index of the calibration accuracy.The genetic algorithm is used to optimize the smallest singular value of identification Jacobian matrix.In the end,this method will explore a suitable calibration board pose combination,and thus the calibration accuracy is improved.In addition,the epipolar geometry principle of the stereo camera is also used to solve the numerical solution of the stereo camera’s epipolar correction parameters,which can be used as the initial value for solving the correction parameters.Secondly,when the calibrated stereo camera is used to realize the driver’s sight tracking,an efficient eye positioning and image matching algorithm ensures the real-time performance of the system.Sobel operator based eye positioning algorithm accurately locates the center pixel coordinates of the eyes in the left image of the stereo camera.The algorithm uses the stereo matching method to obtain the corresponding points of the center pixel coordinates of the eyes in the right image.Then,the coordinates of the eyes center relative to the stereo camera coordinate system will be computed.Then,perspective transformation algorithm adjusts the A-pillar blind zone camera image.During this process,four sets of corresponding points are calculated using the solid geometric,and thus the perspective transformation matrix is calculated.After that,the image will be further adjusted and cut.In addition,to improve the accuracy of the A-pillar blind zone visualization system,this article proposes an object pose measurement method based on a homography matrix.This method accurately determines the three-dimensional coordinates of the A-pillar blind zone display and the eye positioning stereo camera in the world coordinate frame.This greatly improves the display effect of the A-pillar blind zone.Finally,this article gives the reasonable simulation results of the proposed algorithms.These algorithms have been applied in the A-pillar blind zone monitoring and visualization experimental platform.From the experimental results,the precision,realtime performance,and comfort of the system can be simultaneously satisfied.Hence,this project has a rather broad application prospect in improving driving safety in the future.