摘要： 为克服传统的立体视觉传感器标定过程繁琐、不适应现场操作等弱点，提出一种基于本质矩阵求解的双目视觉快速自标定方法。该方法离线标定摄像机的内部参数，通过双目匹配的对应点，线性求解双目视觉的本质矩阵，快速地得到双目视觉传感器的各个外部参数。实验结果表明，该方法在精度上与传统标定方法相当，且标定过程快捷、方便，适合现场标定操作。

关键词: 双目视觉, 自标定, 本质矩阵, 矩阵求解, 外部参数, 视觉传感器

Abstract: Most traditional approaches for stereo sensors calibration often require accurate calibration objects and complex procedures, which limit their capability in out-door work. This paper proposes a quick and flexible method for stereo sensors self-calibration which works on the essential matrix. It gets two cameras’ inner parameters off-line. And equation of the essential matrix calculates through matching the corresponding points in two corresponding pictures linearly. The outer parameters can also be got by calculation. Experimental results show that calibration results can be compared with traditional ones, and calibration procedures are quick and convenient. This method is also suitable for out-door operation.

Key words: binocular vision, self-calibration, essential matrix, matrix solving, outer parameter, vision sensor

中图分类号: 

• TP391.41