Research On Technology Of Capture,Display And Reconstruction Based On Integral Imaging

It is the innate human visual specialties that make people put forward new requirements for visual experience constantly to achieve the goal of ‘restoring the real world’.The increasingly mature two-dimensional display technology can satisfy the human visual demand to the image features such as the color,the resolution and the brightness to a certain extent,but the lack of depth information limits the further development of the technology.As early as one hundred years ago,stereo display technology first came out.Integral imaging,which can provide true three-dimensional(3D)stereo images with full color,continuous parallax and continuous perspective for the viewer without visual fatigue in the absence of the auxiliary and the coherent light source,stands out with the continuous development of the related technologies,and there are promising prospects for development.Great improvements have been acquired in the related theory research and practical application process of integral imaging recently,but there are still many challenges,such as image capture,large screen stereo display,watching characteristic parameters(resolution,angle and depth)and so on.This thesis focuses on the key issues of large-screen 3D display technique,image capture technique,reconstruction technique and the related applications,puts forward the corresponding solutions based on in-depth analysis of relevant theories,and the experimental data and related analysis are given to prove the feasibility of the scheme.The main research content and innovative work are as follows:1.Display method based on the glued lens array in integral imaging.To eliminate the effect of dispersion on display and reduce the difficulty of building the large-size display platform,in this thesis,we systematically analyze the imaging characteristics of the glued lens,establish the optical mapping model between the capturen end and the display end,and discuss the effects of lens spacing,imaging distance and pixel size on display effect.The display platform of integral imaging is built by combining the glued lens array with liquid crystal flat-panel display in the experiment,and the display effect quantitatively proves the effectiveness of the proposed method.Compared with the method of die sinking and processing an integral lens array,the advantages of the proposed method are as follows.Firstly,the glued lens consisting of convex lenses and concave lenses,whose designing and processing are less difficult,corresponds to a simple optical system,and the system has a good effect of eliminating dispersion.In contrast,the integral lens array is the equivalent of a single lens,and the lens of the integral lens array shall be an aspherical mirror,whose designing and processing are very difficult,to eliminate dispersion.Secondly,this method is particularly suitable for building the large-size display platform for integral imaging.As to the cinema-level 3D display requirements,the display method using the integral lens array is faced with many difficulties,such as the processing of abrasive tools of the same size as the cinema screen,and the ransportation,installation and adjustment of the lens array,but in the proposed method,many small lens array units can be produced in advance and then connect them with the project equipment through on-site assembly,which can greatly improve the flexibility of setting up the lens array.2.Two methods of generating elemental image array using discrete capture combined with discrete capture and convergent camera array.In order to capture the elemental image array of a real scene,reduce the hardware cost and solve the depth inversion problem in the display process,in this thesis,we propose two capture methods,one is combining discrete capture and adaptive window interception,and the other is based on the convergent camera array.For the imaging light path of parallel camera array and convergent camera array,which are widely used in reality,we build the optical mapping model between the captured images and the elemental images,and solve the deep inversion problem which is annoying in integral imaging from the implementation mechanism.The imaging principle,realization process,error elimination and display effect of the camera array capture method for real scene are expounded systematically and completely in this part,which can be used as the theoretical basis for the design and construction of the capture platform in integral imaging.These methods can generate the elemental image array of a real scene compared to the computer virtual synthesis capture method,obtain high quality elemental images of large size object,while the capture cost and workload are reduced,compared to the optical capture,and save the step of building the complicated optical model compared with the traditional capture method combined optical collection and computer virtual synthesis.For a camera array system consists of ordinary cameras,these methods can realize the real-time generation and storage of the elemental image array,improving the generation efficiency of the elemental image array(especially video sequence)greatly.3.Method of the colored 3D scene computer virtual reconstruction and the content perception.A 3D scene reconstruction method in integral imaging is proposed in order to enable the viewer to know the content of an elemental image array in the absence of the display platform in integral imaging.First of all,an iterative stereo matching algorithm based on region is proposed on the basis of analyzing the distribution characteristics of homologous pixels in the elemental image array,and the parallax diagram of the elemental image array and its corresponding point cloud data are calculated.Then the point cloud is optimized,and the virtual reconstruction of the colored 3D scene and its content perception are completed finally.This research result can be directly applied to the field of non-contact optical 3D measurement.In this method,the elemental image array contains more stereo information compared with the image 3D reconstruction method based on binocular camera,and a large amount of point cloud data with color information can be obtained with only one shot,improving the efficiency of reconstruction greatly,compared with the 3D reconstruction method based on structural light and optical path difference.