基于共线点的镜头畸变校正方法

为实现未知摄像机参数的镜头畸变校正，提出了一种先标定畸变中心、再标定畸变系数的方法。先在镜头的不同焦距处对靶标成两次像，利用相同靶标点在两幅图像中的相对位置关系求解畸变中心；再根据直线的透视投影不变性，通过变步长的最优化方法搜索畸变系数。模拟实验表明，在靶标点数为25，噪声水平为0.2像素时，畸变中心的平均误差为(0.2243,0.1636)像素，畸变系数误差为0.28%。真实图像实验表明，用该方法得到的畸变中心和畸变系数能够很好地校正图像。该方法不需要标定摄像机的内外部参数，也无需知道直线网格的世界坐标，简便易行。

Lens distortion correction method based on collinear points

To carry out lens distortion correction of unknown camera parameters, a method calibrating distortion center first and distortion coefficient second was proposed. First, with target imaged at two different focal lengths, distortion center could be positioned using relative position relation of the same target points in two pictures; then according to the invariance of line perspective projection, distortion coefficient could be searched by variable step optimization method. The simulation experiment showed that the average error of distortion center was (0.2243,0.1636) pixel, and the distortion coefficient error was 0.28% when target point number was 25, the noise level was 0.2 pixel. Real image experiment shows that distortion center and distortion coefficient obtained by the proposed method can correct the image well. The method is easy to implement, the calibration of camera's internal and external parameters and the knowledge of line grid's world coordinates can be neglected.