Investigation On The3D Information Acquisition Based On The Structured Light

In the machine vision measurement, the3D information acquisition technology based on thestructured light has been becoming an important technology for active non-contact opticalmeasurement, due to its high accuracy, low cost, fast process and easy operation, and gets anextensive development in the fields of industrial inspection, reverse engineering, rapid prototyping,mold, medical, cultural protection, et al. In the principle of optical triangulation, the structured lightmeasurement system generally consists of a camera, a projector and a computer. Projector projectthe structured light on the object, then the camera shot the distortion image and sent to the computerto process, finally, we can get3D shape.This thesis preliminary investigates on the3D information acquisition technology based on thestructured light, mainly contains the following work:Firstly, introduces the related knowledge of the3D information acquisition technology based onthe structured light, and studies the coding method of the structured light.Secondly, calibration is a key step, of which the calibration accuracy directly influences themeasurement results. This paper establishes the mathematical model, the distortion model, planecalibration model, and the relationship between the3D space coordinate and the2D imagecoordinate.Thirdly, given the properties of lens distortion, this paper presents a new calibration method, inwhich the target region is divided into many sub-regions of concentric rings formed by treating theoptical axis as a center point along with a given radius, then, according to the reprojection method,calibrations are done one by one. Since this method takes into account the impact of the large viewas well as lens distortion, calibration accuracy of both local and global regions is improveddramatically. The experimental results show that to the field of about1050mm×750mm dividedinto two regions, the accuracy for projector calibration increases by72.8%and the system precisionreaches28.9μm, indicative of advantage of our method over distortion processing.Fourthly, constructs the mathematical method of3D point cloud acquisition, and obtains thedepth information by image processing, feature extraction and decoding, then completes the3Dscanning and reconstruction.