3D articulated object understanding, learning, and recognition from 2D images

This paper is aimed at 3D object understanding from 2D images, including ar ticulated objects in active vision environment, using interactive, and inte rnet virtual reality techniques. Generally speaking, an articulated object can be divided into two portions: main rigid portion and articulated portio n. It is more complicated that "rigid" object in that the relative position s, shapes or angles between the main portion and the articulated portion ha ve essentially infinite variations, in addition to the infinite variations of each individual rigid portions due to orientations, rotations and topolo gical transformations. A new method generalized from linear combination is employed to investigate such problems. It uses very few learning samples, a nd can describe, understand, and recognize 3D articulated objects while the objects status is being changed in an active vision environment.