一种精确标定摄像机的遗传算法方案(英文)

A GENETIC ALGORITHM APPROACH TO ACCURATE CALIBRATION OF CAMERA 

摄像机标定的一个重要问题是既要保证参数估计的精确性,又要满足旋转矩阵正交约束条件.本文在线性优化和遗传算法的基础上提出了一种新的双段标定方案.第一阶段,在忽略畸变参数的情况下,利用最小二乘法估计摄像机内外参数.为减少畸变影响,本阶段仅考虑图像中心区域的标定点;第二阶段,以第一阶段的估计值为初始值,并考虑摄像机畸变的影响,根据整个图像范围的标定点,利用遗传算法优化摄像机的所有参数以获得精确解.由于该算法直接优化摄像机的旋转角度,所以本文的标定方案能够容易地满足旋转矩阵正交约束条件.此外。

A new two stage calibration method based on the close form solution and genetic algorithm was presented. In the first stage, the distortion parameters were ignored. The internal and external parameters were estimated by least square algorithm. In the second stage, the parameters estimated in the first stage were used as initial values. Taking account of the camera distortions, all parameters were optimized by genetic algorithm to get exact solutions. Because the algorithm optimized the rotation angles directly, the orthonormal constraints were easily satisfied in the present method. Moreover, the approach simplified the calibration process.