Research On Encoding And Decoding Technology Of Structured Light 3D Reconstruction

3D reconstruction refers to the process of establishing mathematical model of space object by means of computer and other auxiliary equipment,it is the base for management,data process and analysis in computer environment,and it is also the key technology of virtual reality.Currently,the acquisition for three-dimensional shape has been widely used in such as computer graphics,virtual reality,medical diagnosis,robot vision,industrial testing,reverse engineering and other fields.Traditional methods for 3D reconstruction,such as passive 3D reconstruction,has the problem of low matching accuracy and the inability to handle low texture objects.Therefore,more and more attention has been paid to the active 3D reconstruction technology,and the main method of active 3D reconstruction is based on structured light.The 3D reconstruction based on structured light usually generate a series of images in advance,and then project these images through the projector and other equipment to the object,by artificially increase the texture information of the object to improve the accuracy of reconstruction.Among them,the combination of gray code and sinusoidal fringe code is the most classic one and it has been widely used,in recent years,more and more people begin to use color coding and decoding.In this thesis,firstly,we study and improve the classical method which combines gray code and phase shift,and then we combine the advantages of traditional method and color encoding and put forward a kind of method which combines gray code and color encoding.The specific content and the innovation work is demonstrated as follows:1.The traditional method of combining the gray code with the phase shift usually first projects a series of gray code fringes to the measured object,determines the approximate area on the surface of the object,and then projects a series of sinusoidal fringes,subdividing the previously determined regions,getting the corresponding information of each pixel on the image,and finally determines the corresponding relationship of the same point on the object between the left and right images.But in the traditional method,when the sinusoidal fringe is projected by a projector and collected by a camera,it will lose some sinusoidal characteristics,in order to solve this problem,we introduce modulation intensity function to compensated the sine grating which is generated by the computer by cyclic iteration method,it can improve the reliability of phase and the final experimental results.2.Traditional black and white fringes contain less information,and color images often contain more information,so in combination with the advantages of these two methods,we put forward a method of combining gray code with color coding.The first step is still to project s series of gray fringes to the measured object,determine the approximate area on the surface of the object,and then project a pre-designed color coding image,find the corresponding point by the idea of traditional local matching.This method can not only avoid the occurrence of periodic inconsistency in the traditional method,but also improve the experimental accuracy of the color coding method,reduce the search range of the traditional matching algorithm,besides it can reduce the number of projected images and has obtained some good experimental results.