The Parameters Research Of 3D Vision Reconstruction System Based On Parallel Mechanism

In recent year, parallel mechanism research becomes a focus field because of its merits. The combination of 6-DOF parallel robot and structured light technique builds a 3D vision reconstruction system based on parallel mechanism, and the system takes on significance and widely application field. This paper is to study the main parameters of 3D vision reconstruction system based on parallel mechanism. We studied some important target of the system on the foundation of an in-deep analysis of status quo at home and abroad.Firstly, this paper schemed out the space light-path, established the math-transform model among space coordinate, camera-observation coordinate and image -pixel coordinate by analyzing the math model of light projection triangle measurement. The model simplified the original model and made the system have general application.Secondly, this paper optimized the design of the system's parameters by analyzing factors such as CCD camera system, digital projection system, parallel mechanism workspace and system error etc.On this basis, given the reconstruction space scope of 3D vision reconstruction system based on parallel mechanism.Thirdly, the study focused on self-occlusion, which is prevailing in structured light 3D vision reconstruction system, and proposed a method of detecting self-occlusion, the method is using a parametric quadric surface based on the Euclidean distance to estimate the point cloud data curvature, by analyzing the changes of point cloud data curvature to judge whether there is self-occlusion on reconstruction object. Experiments indicated that there are many differences of the curvature change between step point cloud and continuous point cloud. Using this particular curvature change phenomenon, we can judge whether there is self-occlusion on the surface of reconstruction object, and more precisely located the position of self-occlusion. Parallel mechanism has the distinct advantage of space location, it can reduce the self-occlusion phenomenon with its characteristics of moving, so, we can obtain more information of 3D object surface.At last, this paper verified the effectiveness and the correctness of system error and self-occlusion algorithm by simulation experiments.