Surface Reconstruction Based on Computer Stereo Vision Using Structured Light Projection

Object reconstruction is one of the most important topics in computer vision due to its wide field of application. Reverse engineering has an increasing need for reconstruction of stereo parts. A system for reconstruction three dimensional (3-D) object's surface from its 2-D views using a coded structured light combining with the epipolar geometry that exists between the stereo image pair is presented. A set of multiple gray coded patterns and a set of sine wave patterns are combined to form a single "gray code composite pattern" and a single "phase shift composite pattern" respectively. These two structured light patterns are projected onto the target object through a DLP projector and the distorted strips patterns produced by its surface are captured by two CCD cameras. By demodulating the distorted patterns and adding epipolar constraints the match points between the image pair are found out and then the target object's 3-D point's coordinates on the surface are obtained. The system shows excellent linearity and the results shows good resolution.