Extracting And Processing Orthographic Projection Information Of Engineering Drawings For 3D Reconstruction

3D reconstruction technology based on engineering drawings is to construct 3D objects by use of geometric and topological information of 2D orthographic projection views. The process of construction is from lower dimension to high dimension, which has become an important research subject in CAD/CAM.As the first step of reconstruction, information extraction and processing plays an important role in 3D reconstruction. The thesis made deeply research to the technology of information extraction and processing in engineering drawings based 3D reconstruction method of characteristic recognition under the background of technological tacking key problem for Liaoning province, which obtains some production. The main contributions are summarized as follows:First, the format of DXF files and storing structure is researched and analyzed detailedly. Interface program to reading primitive date from DXF under CAD system is also designed and programmed. Analying the dimension mark restriction semantics and summarizing the algorithm of extracting dimension mark.Second, on views separation , it presents an multi-views separation method of the least circum-rectangle. It reduces times of travelling meta by setting a sign-bit in programming and implements separation of various nonstandard three views including one view, two views etc..Combining engineering semantics, it confirms the views relation and implements coordinate conversion from drawing reference frame to space reference frame.Third, the thesis emphatically discusses the method of recognition and processing to the half-cutaway views. According to the format of DXF files, the thesis presents and implements the algorithm that extracts the information of half-cutaway views. The algorithm recovers the figure which is cutted by half-cutaway views , which debases the difficulties for 3D reconstruction.To make a summary , the scheme of engineering drawing pre-processing for 3D reconstruction are presented in this thesis from the viewpoint of engineering application. It is more nicety and integrity than traditional method in pre-processing the information of engineering drawings through extracting and recognizing semantics information and half-cutaway view information. The scope of processing object is expanded from standard three views to actual engineering drawings, which expands applicability to 3D reconstruction radically. The algorithms in this thesis have been experimented in 3D reconstruction system and shown to meet the need of 3D reconstruction.