基于交比不变性的广角成像系统几何失真校正

广角成像系统在近距离倾斜摄像获取大场景平面信息时,图像会存在透视变形与径向失真综合的非线性几何失真。利用网格模板成像,结合透视投影下的共线点列的交比不变性原理,通过非线性优化获得径向失真模型参数,实现径向失真的校正;通过透视变换方法构建变换矩阵求得变换系数,进而实现线性梯形失真的校正。对广角镜头焦距为14mm,相机倾斜角为43.34°,边长1200mm正方形模板成像的几何失真进行了校正实验,经对不同位置分布点的精度分析,得到各位置点的相对误差均小于1%。表明校正算法适用于非线性几何失真的处理,且具有较高的校正精度。

Correction of Geometric Distortion of Wide-Angle Imaging System by Cross Ratio Invariability

When using the wide-angle imaging system to obtain large flat-screen scene information close and tilt, images will produce nonlinear geometric distortion consisted with perspective distortion and radial distortion. Based on imaging the grid template and the cross ratio invariability of collinear points with perspective projection, the radial distortion model parameters by nonlinear optimization is obtained, and the radial distortion is corrected. The transform matrix is built to calculate the transform coefficients by perspective transform method and correct the trapezium distortion. The non-linear geometric distortion is corrected for the images of square template with the side length as 1200 mm, the focal length of wide-angle lens as 14 mm, and the camera tilt angle as 43.34°. By the accuracy analysis of points in different locations, the relative error of identification points are less than 1%, the results show that the algorithms are suitable for the correction of nonlinear geometric distortion and have a high correct accuracy.