Camera pose estimation but camera worldOrientation is known
6 views (last 30 days)
I am working on a visual navigation problem. I have a set of aerial imagery and its corresponding satellite imagery. From these two images I am able to extract world coordinates and apply them to my image frame. This information can then be easily fed into estimateWorldCameraPose to get the worldOrientation and Location of the camera.
Since my camera is attached to an aircraft I know exactly what the worldOrientation is of my camera. So my question is there a a way or a function that I can feed the orientation in to speed up and refine my worldLocation estimate of the camera, by esentially reducing the degrees of freedom?
Prabhan Purwar on 28 Jul 2021
poseGraph3D is used to store the poses in terms of edges and nodes that can be optimised using graph optimization for better trajectory estimation. addRelativePose is used to add the poses along with uncertainty representation in terms of the Information matrix. Orientation obtained using aircraft position can be fed with higher certainty to get better pose estimation and trajectory representation.