Research On 3D Modeling Technology Based On Multi-source Measurement Data Fusion

With the rapid development of basic science in various fields and the continuous expansion of information collection and application fields,people are no longer satisfied with expressing three-dimensional information of the real world through two-dimensional data.There are many limitations in the representation of three-dimensional spatial entities,and a large amount of three-dimensional spatial information is lost.On the other hand,its human-computer interaction interface is very inconvenient and it is difficult for people to enter the real world.The rapid development of laser measurement technology has provided a more convenient way for acquiring three-dimensional spatial information.It has become a popular technology in the field of non cooperative targets,such as fast and high-precision 3D ranging and automatic acquisition and transmission in distance and angle.At the same time,the upsurge of concepts such as virtual reality and digital campus has swept over the world,and more and more researchers devote a lot of time and money to their research.Based on the actual project,the research content of this paper mainly divides into four parts as follows:(1)The three-dimensional laser scanner will produce the problem that the accuracy does not meet the requirements of the project in the scanning operation,leading to the following three-dimensional model accuracy can not meet the requirements.This paper systematically analyzes the error sources of the scanner,and uses the unilateral method and the cross-bilateral method to calibrate the experimental program,which can quickly and accurately check the accuracy of the three-dimensional laser scanner,and lays a good foundation for subsequent data acquisition.(2)Traditional texture image collection methods have no rules,there are many pictures taken,the data volume is large,and some scene information is sometimes omitted.Through comparative analysis of a large number of texture image data collected before and after several times,a fast and comprehensive texture acquisition method was proposed,which improved the collection efficiency and reduced the data volume.Through the research of noise reduction,feature enhancement and other algorithms,the texture image is processed,and a good model is shown to be exquisite.Finally,through comparison experiments,theinfluence of three different texture mapping methods on the degree of model authenticity,memory occupancy and ease of operation is analyzed,and the advantages,disadvantages and applicable scope of each mapping method are obtained.It provides a theoretical basis for subsequent high quality,fast texture maps.(3)For the problem that terrestrial laser scanner fails to recognize the target sphere when it is stitched in the point cloud,a large number of other point cloud data and target placement positions are analyzed and studied.Then the corresponding placement rules are proposed to improve the success rate of the recognition target and the efficiency of point cloud splicing.The complex surface class model is difficult to operate in the forward modeling software,and the operation is complicated.The author uses the conversion format to place the point cloud in the reverse software using the surface fitting modeling method to perform three-dimensional modeling and improve modeling effectiveness.Irregular class models will disappear when they are manipulated by intersections,unions and difference sets.After the experiment and research,the modeling steps are proposed in detail,which reduces the occurrence of such problems.The model integration stage of teamwork is prone to problems of material and model duplication.Combined with the modeling technical requirements of the project,relevant modeling modeling specifications are proposed and the efficiency of model integration is improved.(4)One sewage treatment station in the project was selected as an example to study the entire process from data acquisition,data processing to 3D modeling.First,precision calibration of the three-dimensional laser scanner;then,collecting data such as point cloud data,texture image data,and process flow chart data of the sewage treatment station,and splicing and noise reduction processing of the acquired point cloud data,and finishing and processing the texture Image data;Finally,three-dimensional modeling and texture mapping of the wastewater treatment plant.