Research For The Method Of Improving Structured Light Scanning Point Cloud Using Stereo Image Processing

Along with various optical and sensor development,three-dimensional scanningmeasurement has officially entered into all fields， regardless of model processing,CNCmachining,cultural relics protection,construction,3D scaning can be seen everywhere. Thistechnology which can be efficiently measure the target3D information--3D scanning,declaration of Surveying and Mapping Science and technology finally formally entered theera of the three-dimensional measurement.After geodesy, photogrammetry, mapping,engineering surveying, surveying and mapping technology almost created a new area.Currently, there exist a variety of scanner： terrestrial laser scanners, structured light3Dscanner, handheld macro laser scanner and so on. Main defect of the structured lightphotogrammetry which is a kind of very effective way is that the cloud point are lost in edgeof scanned object. This study is to present a method that can remedy the flaw. The structuredlight scanning should be more feasible since its images are the original data accessed. Thisresearch tried to combine the image processing method to a structured light system in order toimprove the quality of the point cloud and the TIN derived. The approach is followed asbelow： Firstly, a high-precision edge extraction algorithm is used to extract the edge fromeach image of a pair of stereo image respectively. Exactly,a Positioning Operator is adopted toposition the point and edge of the image feature by sub-pixel accuracy. Secondly, based on anEpipolar Constraint, the corresponding image features (point and edge) are determined onlyfrom the results of the feature extraction. The3D coordinates of every pair of points arecalculated with the functions of photogrammetry and then, the3D edge model are built. Afteroverlying the edge part of the3D model to the original point cloud from structured lightsystem, their hiatus can be restored and the resolution of the original point cloud can beimproved.To achieve this goal, this paper will take the following methods：(1) Feature extraction using ROG for stereo image edge feature extraction.(2) Stereo matching, based on the strength of the matching algorithm, using a5*5window to create a target feature domain, along the epipolar line to search strategy.(3) Using intersection method is derived to calculate the3D position of points to be.After a series of research and experiments, after scanning, feature extraction, stereomatching,3D solution and other steps, finally calculates the structure light scanning of amissing part of the3D point cloud. This paper proposed the use of digital photogrammetrytechnology to enhance the structure light scanning point cloud3D reconstruction method.Make full use of two kinds of methods of measuring their respective advantages, completed aseries of key technology research, has a certain practical significance.