Research On Technology Of Data Processing Of The Rapidreverse System Based On Deep Image

As the key technology of the Rapid Reverse System data processing consists of pre-processing and surface reconstruction. The result of the data processing will directly influence the precision of the finial model. So modifying the algorithm of the data processing and improving the precision can improve the capability of the system. In this paper we focus on data processing of the Rapid Reverse System. Deep research will be done about the data transfer, data pre-processing and NURBS surface reconstruction. 1. Firstly the paper discusses the problem of data transfer between CMM and CAD/CAM system. Secondly the paper chooses the IGES format as the realizing standard of data transfer, analyze its document structure and on this basis the algorithm of data transfer based on IGES is put forward. Finally the paper designs the corresponding data transfer interface using the algorithm. 2. Aiming at the measuring data points gained from the processing of the corresponding deep image, the data pre-processing, such as data filtering, extraction and segmentation were put forward. A data filter method using the least curve fitting on the basis of adjacent region was proposed. A data extraction method using the combination of the angel and windage was proposed and the method can not only effectively compress the huge amount of redundancy points but can keep the outline of the model more accurately. A data segmentation method base on SOM was proposed. The algorithm was amended by using the quasi-center of clustering and adding right to the input vectors. The method can segment the measuring data points into data blocks based on the corresponding characters, and the data subjected to the same surface was segmented to one block, which is convenience to the subsequent surface reconstruction. 3. NURBS surface reconstruction of measuring data points.The measuring data after pre-processing was fitted with NURBS curve with constructing the non-uniform node vector using the method of cumulating chord length and optimizing the rights of NURBS curve restrictively. The formula of surface skinning which can be realized by programming was proposed on the basis of the foundation formulas in existence, then NURBS curves was skinned to gain the corresponding NURBS surface. 4. A data processing software consists of data pre-processing and surface reconstruction was designed, programmed by Visual C++. The software includes the function modules of data filter, extraction, segmentation and surface reconstruction, furthermore the graphics show and alternation function was realized using OpenGL.