Research On Dynamic Real-time Display Algorithm Of Automotive A-pillar Blind Spot

Traffic accidents occur frequently in modern life,and the blind area of a car is an important factor that affects the driver’s judgment and is one of the important causes of traffic accidents.The A-pillar of a car is the support column between the front windshield and the door of the car,and is an important structure to ensure the strength of the car body and the safety of the driver.But at the same time,it will also block the driver’s field of vision and affect the driver’s judgment.The elimination of the blind area of A-pillar of the car can greatly reduce the related traffic accidents.In light of this,this thesis proposes a blind spot display system that combines image registration and head pose,which can display the corresponding blind spot in real time without changing driving habits.This method requires a lot of data.Therefore,the algorithm is optimized according to the corresponding relationship between head movement and blind spot image boundary movement.Then this thesis proposes a method that uses the center of the pupil to calculate and display the blind zone in real time.Due to the particularity of the installation position of the camera and the blind zone display screen,it is recommended to use the coordinate system conversion method to calculate the three-dimensional coordinates of the spatial point in the non-field of view.Due to the limitations of the existing algorithms for pupil center detection,when the pupil center detection fails,it is proposed to use head posture compensation to ensure the stability of the system.The specific research content is as follows:Firstly,analyze the occlusion of the driver’s vision by the car’s A-pillar in normal driving,and put forward the hypothesis--the blind area generated by the A-pillar is unchanged for the same driver,that is,The relative position of the same driver’s head and A-pillar is fixed.Based on this hypothesis,it is proposed to use two cameras to shoot the A-pillar blind area and panoramic images at the same time,and calculate the A-pillar blind area through the image registration method.At the same time,the camera is used to photograph the head posture at the current position,and the z-axis offset in the head posture is fitted with the blind spot position,thereby realizing different blind spots for different drivers.After that,the relationship between head movement and blind spot image boundary movement was studied,and the algorithm was improved and optimized.The experimental results show that the corresponding A-pillar blind area can be effectively displayed for different offsets.Secondly,analyze the shortcomings of the above algorithm.This paper designs an algorithm that uses the pupil center to calculate the blind area of the A-pillar,and uses the head posture to compensate.The algorithm uses the imaging characteristics of the human eye to calculate the blind zone by simulating the line of sight,and when the pupil center detection fails,using head pose is used to compensate.Experiments show that the algorithm can effectively calculate and display blind areas in real time.Thirdly,since the installation position of the A-pillar camera is above the display screen,it is impossible to directly calculate the coordinates of the corner point of the display screen under the A-pillar camera.In this paper,by using the coordinate system installation method,the corner coordinates of the display screen are converted to the A-pillar camera coordinate system through the auxiliary coordinate system.Experiments show that this method can effectively handle the calculation of spatial point coordinates in non-field of view.