Research On Key Technologies Of Automatic 3D Reconstruction Of CAD Architectural Drawings

With the promotion of smart city and the development of high-tech technologies such as Internet of Things,China is accelerating the process of digital city construction,and 3d reconstruction of buildings is an important part of digital city construction.Among the multiple data sources available for building 3D reconstruction,architectural drawings are the most widely used and contain the most detailed information,which has unique advantages in building interior reconstruction and high-precision restoration.Compared with traditional manual image recognition modeling,automatic 3D reconstruction has low cost and high efficiency,and has gradually become a research hotspot.The research content includes automatic recognition and automatic modeling two key technologies.Among them,automatic recognition takes semantic recognition of building components as the core,and corresponding recognition algorithms are designed for all kinds of building components to extract target location and contour information.However,due to the lack of in-depth mining of drawing information and insufficient general applicability of the algorithm,it is difficult to meet the needs of high-precision reconstruction of complex building drawings.In the 3d engine,the 2d information extracted from the recognition results of drawings is converted into 3d models according to certain rules.However,when the data volume of complex model of drawings is too large,the visualization effect is not good,and it is difficult to make a choice between model simplification and detail retention.Aiming at the above problems,this paper studies the key technologies of automatic 3D reconstruction of complex CAD architectural drawings.The main work includes:(1)Refined identification method of complex architectural drawingsIn view of complex architectural drawings,taking doors,Windows and floor components as the research object,the recognition method of component subclassification and special structure is explored.Based on the recognition rules of ordinary doors and Windows,the recognition rules of more subcategories are established,and the recognition algorithm is written by taking civil air defense doors,sliding doors and bay Windows as examples.On the basis of recognizing the outer contour of the floor,an algorithm is designed to recognize the hollow well with special structure inside the floor.(2)Automatic building modeling and lightweight methodsBased on automatic identification results,in the 3 d engine according to certain rules algorithm automatically modeling,real-time rendering,the problem of the large amount of data and model designs a nondestructive lightweight methods,combining with the construction characteristics,the use of adjacency between the construction and the similarity between the multi-storey building floor,wall column,for example in geometry layer shall merge optimization,eliminate redundant information.(3)Automatic 3D reconstruction system construction and experimental verificationBased on object Arx and unity3 D,the automatic 3D reconstruction system of CAD architectural drawings is developed.The system integrates the CAD plug-in module with the functions of loading,reading and automatic recognition and the 3D engine module with the functions of automatic modeling and lightweight,realizing the automatic reconstruction of the whole process of complex architectural drawings.In this system,the validity of the research content is verified according to the scientific index.The experimental results show that the fine recognition method proposed in this paper and the automatic modeling lightweight method can efficiently complete automatic 3 d reconstruction of complex architectural drawings,model precision degree is high,3 d visual effect is good,can play a certain guidance effect for design and construction personnel,for the construction of high precision digital city construction provides effective means.