Research On Handy Self-localization Additive Manufacturing Morphology Detection Technology

AM is a technology of making components by stacking layer and layer.Since there are many uncontrollable variables,such as heat dissipation conditions,heat accumulation and other factors,the stacking size is often inaccurate.Therefore,how to increase the precision of AM products is one of the main research problems.In this paper,a set of three-dimensional scanning system with hand-held and self-positioning features has been developed,and a closed-loop AM system has been established to improve the dimensional accuracy of the products.First,the paper has created a handy and self-localization three-dimensional scanning experiment system.The internal and external parameters of the cameras are calibrated by Zhang calibration method,and the paper proposed an automatically and efficiently calibrate method for linear laser plane equation.The three-dimensional coordinate of calibration point is calculated with the binocular vision principles,and the accuracy is under 0.2 mm.According to the laser plane equation,the three-dimensional information of single-frame image is restored,and the accuracy is under 0.1 mm.The algorithm of central coordinate extraction for circular calibration point is optimized.The matching of the points in different camera images of same frame and the points in different frames is realized.Using the calibration points information the paper puts different frame image data together and gets the whole points cloud.According to the above principles,the paper uses C++ language to achieve a software with online image processing and displaying the three-dimensional point cloud functions.The scanning accuracy is under 0.5 mm,the scanning speed is about 15 cm per minute in the case that the maximum distance is under 0.5 mm.As the arc additive manufacturing is generally carried out on the plate,this paper proposes an algorithm to uniform point cloud,which effectively eliminates the repeated information and greatly reduces the number of point clouds.The algorithm improves the efficiency of point cloud processing.With point cloud statistic denosing,Gaussian filtering and greedy projection triangulation algorithms which is based on the Delaunay triangular,the paper has rebuild the point cloud into a smooth surface.The paper has developed a software with C++ language to achieve the functions.The software can display the dynamic 3D model n the MFC environment with Open GL 3D display technology.The 3D model is saved in STL format,leaving the interface for other 3D model processing software such as CAD,Solid Works.Finally,the feature information extraction algorithm of different stacking situation is proposed.We rebuild each stacking layer three-dimensional morphology and extract the feature information.According to the feature information,the next stacking parameters and the paths are improved.The closed-loop stacking system is established to improve the forming quality of the AM products.