一种成像测量图像径向几何畸变的校正方法

Radial geometrical distortion correction in image measurement system

通过对具有径向畸变的摄像机模型的分析,设计了一套求解图像径向几何畸变中心和畸变多项式系数的方案。首先,依据校正样板曲线的弯曲程度应用一元线性回归法和逐次逼近法求取光学图像的几何畸变中心,然后应用递推最小二乘法求解径向几何畸变的多项式系数,最后根据所得到的畸变中心和畸变多项式系数对图像进行校正得到满足要求的图像。仿真试验证明:该方法可以通过一次采集单幅图像对成像系统进行高精度标定,能够对成像测量系统的径向几何畸变进行一定精度的校正。实践证明:该方法通过图像处理的方法提高成像测量系统的精度,降低了系统的设计成本,可以作为成像测量系统中单独标定摄像机畸变参数的一种简单有效的方法。

Lens distortion in high precision imaging measurement system has to be strictly controlled,though the design and manufacture of this kind of lens are challenging.Thanks to the maturity of digital technical,image processing offers a new solution to correct the nonlinear distortion of lens.A set of simple and effective methods for detecting radial distortion center and calculating polynomial coefficients was proposed based on the analysis of a nonlinear camera model.Unitary linear regression analysis and gradually approach method were used to obtain the distortion center and recursive least squares estimation was used to get the polynomial coefficients,and images were corrected based on these data for fulfilling requirements.Test and simulation show the radial geometrical distortion can be accurately corrected in imaging measurement system by sampling one frame of image.It is simple and cost effective,could be used to calibrate camera distortion.