Research On 3D Imaging Technology Based On Structured Light

In recent years,with the comprehensive implementation of “intelligent manufacturing” in our country,three-dimensional imaging technology is increasingly important in the fields of industrial detection and industrial design.Recently,the concept of “Metaverse” was introduced,which has stimulated widespread discussion in the society.The so-called “Metaverse” is actually a virtual network world that works on three-dimensional display,and three-dimensional imaging technology provides the important three-dimensional content source to build the “Metaverse”.Threedimensional imaging is fully alternative to two-dimensional imaging and has become a trend for future technology development.We focus on the structured light threedimensional imaging technique,which has advantages of simplicity in hardware,low cost,low algorithm complexity,high imaging accuracy,and fast imaging speed,and plays a key role in the field of three-dimensional imaging.In this thesis,the principle of structured light three-dimensional imaging technology has been examined in depth,including the fringe projection profilometry,wrapped phase calculation,phase unwrapping,system model establishment,system calibration,three-dimensional reconstruction,etc.Specific details of the technology were discussed exhaustively,and a monocular structured light system was constructed,where the core components included an industrial camera and a projector.A series of experiments for system calibration and 3D imaging were further designed and completed.The specific research content includes the following main aspects:（1）The history and research status of the structured light three-dimensional imaging technology were reviewed thoroughly.Based on the high-precision timeoverlapping cosine fringe pattern projection phase profilometry,the basic principle of the phase-shifting method was explored,and the calculation formula of the wrapped phase was derived.To solve the problem of ambiguous phases derived from the highfrequency fringe images,this thesis analyzed the principles,advantages and disadvantages of three state-of-the-art phase unwrapping algorithms,and performed detailed simulation experiments.（2）The mathematical model of the structured light system was established based on the imaging principle of the camera and the display characteristics of the projector.On this basis,the principle and implementation flow of the system calibration were analyzed.A calibration board suitable for use in this thesis was designed and customized,and the calibration of the internal and external parameters of the system has been completed.（3）The camera and projector with optimal performance parameters were selected according to the desired three-dimensional imaging requirements to build the structured light imaging system.And the synchronization of the camera image capturing action and the projector image projecting action was established.Based on the theoretical research of the phase profilometry,three-dimensional imaging experiments were designed and completed using the proposed structured light imaging system.We demonstrated that the system has achieved good imaging accuracy,and can meet the needs of practical applications.（4）An end-to-end deep convolutional neural network model was designed to infer the scene depth information from a single projected fringe image directly.Based on computer graphics,a three-dimensional virtual reality imaging dataset was constructed in Blender,a three-dimensional scene creation software,to assist in training the model.Finally,the model performance on real systems was tested,which proved the feasibility of the proposed method.