靶场光学镜头畸变校正方法研究

摘要

One of the key factors which influence the accuracy of the optical devices in a range is the distortion of the camera lens, especially when the camera Field of View(FOV) is very wide under the condition of short focal length.To some extent, current calibration methods works well.However, the residual errors are still so large that cannot be neglected.In order to eliminate the effect of these errors and improve the angle measuring accuracy, we propose two algorithms called Local Least Square (LLS) and Fractal Interpolation Surface(FIS) to calibrate the lens.Firstly, the FOV(40°×30°) is meshed by an interval of 1° and the optical distortion values of the grid points are measured with precise collimator.In addition, the LLS utilize N×N grid values around the measuring point as observations to fit the calibration function and to find optimal solutions of the parameters of the function.Then the value of the measuring points will be calibrated by this fit function.On the other hand, the main idea of the FIS is to calibrate the measuring points by a fractal interpolation of all values of the grid.The experiments show that both the local optimization of LLS and the global interpolation of FIS can improve the angle measuring accuracy of the optical camera.The accuracy can achieve a magnitude of 10″which can well fulfill the testing requirement of the range and have a positive effect on final location positioning.%光学镜头畸变是影响靶场光学设备测试精度的一个重要因素,特别是短焦大视场情况下尤为显著.目前采用的修正方法虽然能起到一定的效果,但是其修正后的残留误差依然较大,不容忽视,为了消除其影响,提高设备测角精度,提出了局部区域最小二乘法和分形插值曲面法,并采用上述两种算法对光学镜头畸变校正进行研究.首先,将整个视场(40°×30°)以网格的方式进行划分,间隔1°,利用平行光管精确测量网格点的畸变值.局部区域最小二乘法是将测量点周围N×N网格点的畸变值作为观测值,组成系数矩阵,求取修正函数的系数,进而求出测量点位置的修正值;分形插值曲面法是将视场内所有网格点数据进行插值,获取测量点位置的修正值.算例结果表明:两种算法分别以局部和全局的方式进行校正,测角精度有明显提高,修正后精度可达10″,满足靶场试验的测试要求,对提高终点坐标测量精度起到了积极的作用.