Research On Multi-view 3D Reconstruction Technology Based On Encoded Structured Light

With the rapid development of modern computer vision technology,optical measurement of three-dimensional object surface is becoming more and more easy and fast.As one of the key technologies of augmented virtual reality,3D reconstruction technology has important practical significance in the fields of industrial measurement,medical imaging,film and television animation and machine vision.Among them,the active 3D reconstruction technology based on structured light has the high precision,good real-time performance and strong anti-interference ability.Therefore,3D reconstruction using structured light method has always been a dynamic field,and people are more and more interested in its development and application.In view of the current research status of 3D reconstruction technology,and combined with the actual application requirements,this thesis designs a multi-view 3D reconstruction system based on Gray code and phase-shift,which is based on coding structured light technology and multi-view stitching strategy.A complete 3D reconstruction process of the system is as follows: Firstly,the Gray code and phase-shift coding pattern to be projected by the projector are designed according to the complexity of the scanned object.Then the target model is driven by the rotary table to rotate in the specified direction,and images in different fixed viewing angles are synchronously captured by the camera.Next,the deformed two-dimensional image which is modulated by the surface height of the target model is transmitted to the computer for decoding processing to obtain depth information.Finally,the nearby 3D reconstruction of single scanned target model is realized after splicing,registration,fusion and segmentation of point cloud data.The main work of this thesis includes:(1)The design of coding and decoding strategy of the system,solve the positive and negative gray code values of phase shift absolute principal phase bits and gray codes;(2)The spatial modeling and calibration experiments of the system are carried out to obtain the spatial transformation parameters and calibration parameters of the system;(3)Hardware selection and software design of the system,which reduce human intervention,improve the accuracy and efficiency of scanning;(4)Point cloud data acquisition and processing,adopting segmented point cloud splicing strategy and improving traditional ICP algorithm to set the normal vector angle threshold to improve the registration accuracy,and to achieve the three-dimensional reconstruction system of a single object;(5)The experimental results of the system are analyzed..By using the idea of control variables and comparative design,it is verified that the multi view 3D reconstruction system based on Gray code and phase-shift in this paper can obtain the reconstruction results with the error range of 0.6%-3%,and the average reconstruction speed is 43.48 s,which is suitable for the3 D reconstruction of most object models.