Research On 3D Visual Rendering Technology Of Building Model

The development of 3D model visualization technology has greatly improved the use efficiency of model data.With the continuous expansion of social demand,the application of this technology in traditional buildings has become more and more extensive.Building models usually have the characteristics of large data and complex spatial structure,which makes the rendering technology of 3D model data face more difficulties.In order to improve the picking efficiency of grid objects in the 3D building model and enhance the effect of real-time interaction,a ray picking algorithm based on grid space division is proposed.The algorithm uses the hierarchical bounding volume technology as an optimization method,and uses the intersection of rays and triangles as the bottom-level intersection determination algorithm.By subdividing the bounding box space of the target object grid,the number of geometric intersections is reduced and the picking speed is improved.This algorithm was compared with the other three ray pickup algorithms in the experiment.Four different ray paths were selected under two different building model scenes with different data volumes for testing and experimental data were recorded.The performance differences and scene adaptability of the algorithms were analyzed according to the experimental data.In order to effectively optimize the rendering effect of the building model under the dynamic perspective,a frustum culling algorithm based on geometric transformation consistency is proposed.The algorithm uses the method of extracting the frustum plane from the matrix as the detection basis,uses the closest point from the bounding box to the plane as the detection object,and combines the characteristics of the frustum in the dynamic geometric transformation to speed up the re-detection of the object that has been eliminated in the initial state.The experiment uses ten geometric transformation operations to compare the algorithm with the frustum culling algorithm using plane detection on a scene containing one million bounding boxes,and analyzes the performance difference of the two algorithms through the data recorded in the experiment.The experimental results show that the ray picking algorithm based on grid space division effectively improves the picking speed,and can play a better effect in the scene where the model sub-grid itself has a large number of triangles.The frustum culling algorithm based on the consistency of geometric transformation also improves the efficiency of culling detection during the dynamic transformation of the frustum.