Study On Key Techniques For Automatic 3D Structured-Light Scanning System

Structured light scanning method, which based on laser tribulation measurement model, in order to obtain 3D coordinate of the point on the measured object surface, is a promising technique which has been widely used. In this dissertation, a structured light vision system including two cameras and a four axes motion control equipment is proposed to reconstruct the measured object surface integrally and accurately. The major contribution and innovations of this dissertation are as follows:1. Several 3D measurement techniques as well as the instrument at home and abroad are analyzed. Major problems in structured light measurement technique including measurement model, calibration method, image processing and viewpoint automatic selection method are reviewed as well.2. Measurement system combined with a four axes control system and a structured light sensor is built, with two scanning modes translation scanning and rotation scanning according. Measurement model and calculation method are studied as well based on the measurement principle of structured light sensor and position control model of the four axes scanning system.3. Using a chessboard pattern, parameters of both cameras are calibrated out by Zhang calibration method. The light plane parameters are calibrated by using viewpoint uncertainty method to ensure both cameras in a unified world coordinate system. Based on binocular vision principle, the exterior orientation of the turntable's axis is calculated out with a standard ball, employed with axis calibration model and the method of space intersection of two straight lines.4. According to the characteristic of the stripe image, centric line of straight structured light stripe is extracted based on gray level gradient, points filter and improved barycenter method. The sub-pixel center of light stripe is found with high precision and high robust.5. With the four axis motion system, a three-step scanning method is used to determine the scanning path automatically. By putting an unknown object in the measurement volume, the surface data as well as the scanning path is obtained integrally and accurately using the next best view algorithm with the first approximate scanning, the second precise scanning and the third compensate scanning separately.