Research On Building Plans Interpretation And3D Reconstruction

In the visual images’ field, traditional2D images are being replaced by3D images gradually for the reason of people’s visual sense request advancing. People works on3D model manufacture all the time, then they are applied to film-making, machinery, medical treatment and so on. Since the last century, more and more people have worked on the3D reconstruction of buildings. However, the traditional3D reconstruction is mainly dependent on manual, which means only the trained people can finish this work. As the reason of3D construction’s complexity, as well as3D technology has been applied in many fields, automatic3D reconstruction is more and more significant.3D Reconstruction means construct the3D model form the2D images with least manual operate, even completely automatically. This paper is a key technology of project "3D Visualization of Builds’ Fire Extinguishing System and Firemen’s Position Network". The project’s purpose is to detect the fire disaster and forecast the rate of fire spread, and also the accurate position of firemen. Then reflect all above informations in the3D model, to be used for leading the fire fighting. Paper analysis the figure informations, adopt the way of pattern recognition and basics of computer graphics, identify the2D drawings’informations and rebuild3D informations, constructed the whole3D model.This paper presents a set of3D reconstruction program via the research of buildings’ DXF fill, which are base on abundance of reading related papers and perform experiments. This program mainly contains two sections:constructional element recognition and3D construction. As the result of different drawing habits, there are mainly two methods in constructional element recognition:one is drawing straight lint in the way of LINE; the other is drawing a group of parallel lines in the way of MLINE. Building drawings draw in these two different ways looks same, but they are totally different in the DXF information express. This paper comes up with two different constructional element recognition methods to the question of the two different drawing ways.After the constructional elements have been identified, we build the3D model using OpenGL. The model looks more real and well after the lighting and texture handle. Besides, add in the roll function, scale function and move function, we can watch the model in any direction and position, reaching a relatively satisfying result.