In SfM, we would like to recover both the poses of cameras and the position of 3D feature points. We have just seen how simple 2D pair matches of points can help us estimate the essential matrix and thus encode the rigid geometric relationship between views: 
. The essential matrix can be decomposed into 
and 
by way of SVD, and having found and , we proceed with finding the 3D points and fulfilling the SfM task for ...
Stereo reconstruction and SfM
Get Mastering OpenCV 4 - Third Edition now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
