The following are 2 reasons why monocular visual odometry is not the best option:
1) Results from monocular sequences can only be recovered up to a 'scale'; without additional information, absolute measurements are not possible.[src](http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/CROSSLEY1/research/temporal_stereo.html)
2) All pose estimates from a mono VO algorithm are relative to some unknown 'scaling factor'.[src](http://wiki.ros.org/viso2_ros?distro=indigo)
Can someone please explain what exactly is meant by the 'scaling factor' and how does it affect visual odometry?
Thanks
↧
