Study On Key Techniques For Color 3D Laser Scanning System

Color 3D acquisition is a promising technique, which can be widely applied in the fields of reverse engineering, game development, animation, antique digitization, etc. In this dissertation, a color 3D acquisition system based on line structured-light scanning measurement is developed to accurately acquire complete color 3D information, with accuracy higher than 0.1mm. The major contributions and innovations of the dissertation are as follows:1. Various 3D measurement techniques are deeply analyzed and 3D measurement instruments at home and abroad are introduced in detail. Current research work on relevant key techniques, including parameters calibration, color rendering, incomplete-data repairing and NBV planning, are reviewed as well.2. The mathematic model for line structured-light scanning measurement is built, and sensor's unknown parameters are calibrated. With a known-radius standard ball, the position of rotary stage's axis is calibrated. The error due to the fact that only partial spherical surface points data are employed in ball center calculation is compensated, consequently, the calibration accuracy is obviously improved. The system is capable of acquiring the color information of the point-cloud data through repeat-scanning and texture-spread method respectively.3. A novel light-plane calibration technique for double-CCD line structured-light sensor is proposed. A pattern ruled with parallel lines is specially designed to realize the calibration calculation on automatically acquired pixels and world coordinates of control points through image processing. The light-plane parameters are successfully calibrated. Moreover, the measurement references of both two cameras are unified into a common world coordinate system accurately.4. Combining shape from silhouette measurement to acquire the incomplete data of the structured-light scanning. An improved line-segmentation algorithm based on shape-carving is applied to acquire the 3D shape data. And the shape data is used to repair the incomplete data of the structured-light scanning. The completed color 3D data of several kinds of models are achieved after data-repairing though shape from silhouette.5. A solution to next best view planning is proposed. Convex hull is adopted toapproximately express the measured object's shape. Using the normal and location of hull's side as a guide, the sensor's best viewpoint of scanning measurement and color image acquisition is determined automatically. Consequently the system can acquire the color 3D data through automation scanning and color rendering.