Three-dimensional vision from a multisensing mechanism.

An approach for sensing a three-dimensional (3D) object surface with an arbitrary geometric shape is presented. Combining two different 3D sensing mechanisms, point-array encoding based on affine transformation and fringe encoding based on phase mapping, we construct a mathematic model for 3D vision in which the point-array encoding is initially applied to determine the fringe orders to create a control-vertex mesh with absolute coordinate values in 3D space. Then phase evaluation and phase unwrapping for fringe decoding is performed under the guidance of control-vertex mesh, leading to an absolute phase map and the corresponding range image of the test object. The computer simulations and experimental results are presented to demonstrate the theoretical prediction.