该代码使用RANSAC框架中的五点解算器来计算本质矩阵的稳健初始估计。
This code uses five point solvers in a RANSAC framework to compute a robust initial estimate of the essential matrix.
随后,通过使用6个参数(3个用于Rodrigues向量,3个用于平移向量)对本质矩阵进行参数化,并使用Levenberg–Marquardt算法最小化RANSAC inliers距外极线的累积对称距离,从而改进该矩阵的估计结果。
That estimate is subsequently refined by parameterizing the essential matrix with six parameters (3 for the Rodrigues vector and 3 for the translation vector) and minimizing the cumulative symmetric distance from epipolar lines for RANSAC inliers with the Levenberg–Marquardt algorithm.
注意:本代码中使用了多个其它的函数,请阅读README.txt。
NOTE: The code requires several functions by others, see README.txt for further instructions.
关于本质矩阵概念的说明:
See also https://en.wikipedia.org/wiki/Essential_matrix
更多精彩文章请关注公众号: