3D object recognition with image gradient
Dong Xu, Wenli Xu · 2004
The angle between image gradient and the major principal axis of silhouette is proposed to describe 3D object, and the invariance of this description to the density, the orientation of light and the reflection of surface is also deduced in this paper. Using the moment invariants of this description can introduce more 3D information of object and bring higher recognition ratios than using the 2D moment invariant of object silhouettes.