Research On Calibration And Distortion Correction Technique Of Binocular Vision System

With the rapid development of science and technology,people's requirement for the measurement accuracy of objects is increasing day by day.Due to the high accuracy and stability of binocular vision detection technology and the improvement of its demand in various manufacturing fields,this detection technology has become one of the fastest growing technologies.Because the image is affected by many factors in the process of acquisition and transmission,distortion and proportion imbalance occur between the image and the ideal image.Camera calibration and distortion correction technology are the prerequisites of the vision measurement system.The stability and accuracy of the algorithm directly determine the measurement accuracy.Only the real image can reflect the real shape of the object.This paper mainly studies the camera calibration and distortion correction technology of binocular stereo vision measurement system.By analyzing the binocular visual digital model and the principle of three-dimensional measurement,a binocular stereo vision measurement system based on time domain coding is constructed.The image is enhanced by the edge-preserving filtering algorithm based on mean square error,and the image enhancement is realized by histogram equalization.The FAST algorithm based on accelerated segmentation detection is proposed for feature extraction.Aiming at the camera calibration problem,a binocular vision calibration method based on coplanar intersecting circle is proposed.By extracting and solving the circle point coordinates of the elliptical contour of the intersecting circle,the internal parameters of the two cameras and their positional relationship are obtained.Compared with the traditional calibration algorithm,no manual intervention is required and the precision is high.Aiming at the distortion correction processing,it is proposed to convert the camera calibration into a global optimization problem,and the sparse beam adjustment algorithm based on the constraint is used to optimize the parameters.After image preprocessing,camera calibration and distortion correction,theexperimental results show that the average absolute relative error of the parameters in the two cameras is less than 3.45e^-6,the average re-projection error is about 0.1pixels,and the data accuracy is high,which proves the feasibility and theoretical correctness of the experimental method of the system.