unit of translation result from estrelpose function

Hi everyone,

Recently i have tried the example for https://se.mathworks.com/help/vision/ug/monocular-visual-odometry.html, and i used in my own case. But, I ended up with confusion about the unit of translation that produced by estrelpose function in helperEstimateRelativePose. Could anyone give me clue about that? because i try to search until rigidform3d type, but can not find a clear explanation about this. Thank you

Answer 1

[Qu Cao]

As the documentation of estrelpose says, the function calculates the camera location up to an unknow scale. This is becuase you are estimating the relative pose from two monocular images which cannot give you the actual dimension. The norm of the translation should be equal to 1, i.e, norm(relativePose.Translation)=1.