Function mola::vision::decomposeEssentialMatrix

Function Documentation

bool mola::vision::decomposeEssentialMatrix(const Eigen::Matrix3f &E, const std::vector<mrpt::math::TPoint2Df> &pts1_norm, const std::vector<mrpt::math::TPoint2Df> &pts2_norm, Eigen::Matrix3f &R_out, Eigen::Vector3f &t_out)

Decompose essential matrix E into the four candidate (R, t) solutions.

Uses SVD: E = U·Σ·Vᵀ → two rotation candidates × two translation signs. Selects the physically correct one via chirality check (most inlier 3D points in front of both cameras) using the provided normalized correspondences.

Returns the recovered rotation R and unit-scale translation t (‖t‖=1). Returns false if fewer than 10 points pass the chirality test.