On-board Camera Adaptive Exposure And Correction Distortion

On-board camera is commonly used for reversing images,lane departure and blind zone monitoring systems.Due to the particularity of the application background,the image quality of on-board camera has more stringent requirements that it should not only adapt to extreme weather,high dynamic scenes,but also meet the accuracy of the latter part of the test requirements.Based on the functional requirements of blind spot detection system,,this paper studies the lens’ s exposure control and distortion correction to ensure the accuracy of the follow-up target monitoring of the system.In this paper,a self-adaptive exposure control algorithm based on PID and an exposure control algorithm based on image segmentation are put forward to adjust the speed of exposure and to deal with the problem of slow algorithm speed as well as oscillation-prone situation happening in the existing exposure control algorithm.Experimental study indicates,When the PID control method is used for the vehicle lens adaptive exposure which is based on controlling the exposure time,better exposure effect can be got.However,the algorithm base on the global image and has some limitations,it can’t be used in a scene with a change in the range of light.Therefore,this paper proposes an exposure control algorithm based on image partitioning,which uses five-partitioning method and assigns different weights to each region.After weight matrix template processing,it can calculate the problem-oriented exposure value.The results show that the target recognition rate of the PID-based algorithm is 73.1%,and 78.9% is that based on the image-based exposure control algorithm.Therefore,the image partitioning algorithm is superior to the PID control algorithm in that the former can better deal with the regions of interest,avoid infection of other regions,and improve the accuracy of target recognition.The paper analyze the common template method and the known method of parameter analysis,Eventually,propose a new image distortion correction algorithm.Then,this paper presents a new image distortion correction algorithm,which uses the feature point extraction algorithm to get the actual coordinates and the ideal coordinates.On the basis of the above data,a mathematical model is established and the least squares method is used to fit the distortion coefficients.Thus,the distortion coefficients such as centrifugal distortion,radial distortion and tangential distortion are obtained to complete the distortion correction of all the pixels in the image.Finally,a tri-linear interpolation is used to interpolate the pixels of the image,determine the gray level of the pixels,and complete the distortion correction.In the experiment,the image accuracy of the algorithm is verified by the images captured by the vehicle-mounted lens,and the distortion correction algorithm get evaluated by the correction algorithm MRE(maximum residual)and ARE(average residual).The experimental results indicate that the new algorithm of image distortion correction is better than the traditional one in that it can complete image distortion correction more quickly,and achieve targets of higher efficiency,higher speed and higher reliability.