Research On Non-Metric Correction Algorithm Of Camera Lens Distortion

Lens distortion is an important problem in computer vision,photogrammetry,image processing,and other research fields.Due to the manufacturing error of the lens,the lens will have imaging nonlinearity,resulting in geometric distortion.The lens distortion is related to the position of the image coordinates on the imaging plane,which will interfere with the ideal imaging relationship,leading to the actual image point position deviating from the ideal image point position.Therefore,to ensure the effectiveness of subsequent image processing and analysis and maintain the linear imaging process in projective geometry,lens distortion must be corrected.At present,the lens distortion correction algorithm is mainly divided into the metric method and the non-metric method.The measurement method requires precise positioning of feature points in space and estimation of distortion parameters and camera parameters,but these methods have many parameters,a large amount of calculation,and a high correction cost.In contrast,the non-metric method makes full use of the geometric invariants of the ideal pinhole model,such as straight lines,cross ratios,vanishing points,and plane constraints.Therefore,this paper mainly focuses on the non-metric correction algorithm of camera lens distortion.The main contributions and innovations are as follows:(1)Non-metric methods make full use of the geometric invariants of the ideal pinhole model.These geometric invariants represent the properties of the structure and remain unchanged under appropriate transformations.Therefore,these methods do not require some known structural references.In non-metric methods,the straight line is the most widely used geometric feature.Based on the theory of straightness,a non-metric lens distortion correction algorithm based on the geometric characteristics of straightness is proposed.The algorithm uses straightness to define the distortion function and uses the optimization algorithm to calculate the distortion coefficients to complete the distortion correction.In addition,the corner detection algorithm is used to accurately extract the coordinate points in the image,which lays a foundation for subsequent distortion correction.The experimental results show that the algorithm has small computation and an obvious correction effect.(2)The existing straightness method only considers the linear characteristics and does not consider the vanishing point constraint theory.To improve the correction accuracy of lens distortion,a non-metric correction algorithm for lens distortion based on joint measurement of dual characteristics is proposed.The algorithm uses the dual characteristics of vanishing point and straightness to jointly measure lens distortion.In addition,in past work,the image center was always assumed to be the distortion center.In fact,the position of the distortion center will shift due to the offset between the lens center and the CCD center,the slight tilt of the sensor plane relative to the lens,and other factors.To solve the coordinates of the distortion center,the fundamental matrix and the least-square method are used to calculate accurately.The experimental results show that the algorithm has strong applicability and accurate distortion correction effect,and can complete the high-precision correction of lens distortion according to the coordinate points.(3)Based on geometric knowledge,if the vector is collinear and in the same direction,the angle is 0°.Based on the vanishing point constraint theory,a non-metric lens distortion correction algorithm based on the vanishing point collinear constraint is proposed.The vanishing point constraint theory and the collinear constraint theory are used to measure lens distortion.At the same time,the vanishing point is used to constrain the straight line to realize the effective measurement of the lens distortion.In addition,due to process constraints,the pixels after camera discretization are not rectangles but parallelograms.Aiming at this problem,the algorithm constructs a mapping matrix to minimize the pixel error in the horizontal and vertical directions,thereby establishing a more complete distortion model.The experimental results show that the algorithm has a more accurate distortion correction effect and stronger robustness.The processing of pixel units provides a new idea and method for high-precision correction of lens distortion.