光学学报, 2015, 35 (6): 0615001, 网络出版: 2015-05-28
基于直线特征的摄像机镜头畸变标定方法
Camera Distortion Calibration Method Based on Straight Line Characteristics
机器视觉 畸变标定 直线特征 径向畸变 视场角 machine vision distortion correction linear characteristic radial distortion field angle
摘要
在多视域广角成像系统中,为了保证后续图像拼接的质量,必须对从广角镜头所获得的畸变图像进行校正。利用图像中直线特征的畸变标定方法,并且提出一种带权重因子的弯曲测度指标函数,离图像中心不同距离的曲线给予不同的权重值,作为求取最终畸变参数的目标函数。在最小化目标函数过程中,求解出最优畸变系数。并基于工程实例,分别利用传统基于直线特征的畸变校正方法与本文方法对目标图像进行畸变校正。实验结果表明,本文方法仅仅利用单幅图像就能获得高精度的畸变标定,在噪声水平小于2 pixel时,对应坐标的均方根误差能控制在0.3 pixel以内。同时该方法操作简单、方便、易于实现。
Abstract
In the multi-view and wide-angle imaging system, in order to guarantee the quality of subsequent image mosaic, the distorted image obtained from the wide-angle camera must correct. The linear characteristics in the image is utilized to correct the distortion and a bending measurement function is put forward with weight factor that gives each curve line the different weight values in accordance with the distance from the image center as the object function for obtaining the final distortion cofficients. The optimal distortion coefficients is obtained by minimizing the objective function. And based on the engineering examples, are used respectively the traditional distortion correction method based on linear characteristics and the proposed method to correct the target distort image. Experimental results show that the proposed method can only use single image to obtain high accuracy distortion calibration coefficient, under the condition that the noise level is less than 2 pixel, the root mean square error of the corresponding coordinates within 0.3 pixel.At the same time, the proposed method is simple, convenient and easy to implement.
张敏, 金龙旭, 李国宁, 武奕楠, 韩双丽. 基于直线特征的摄像机镜头畸变标定方法[J]. 光学学报, 2015, 35(6): 0615001. Zhang Min, Jin Longxu, Li Guoning, Wu Yinan, Han Shuangli. Camera Distortion Calibration Method Based on Straight Line Characteristics[J]. Acta Optica Sinica, 2015, 35(6): 0615001.