Generation Of Elemental Image Array Based On Spherical Camera Array

Integral imaging technology is one of the hotspots with great development prospects research in true three-dimensional display technology.Compared with other stereoscopic display technologies,the biggest advantage of integral imaging technology is that it can directly see the three-dimensional effect through the human eye without the aid of other auxiliary equipment.The integral imaging display system mainly includes acquisition system and display system.According to the different acquisition methods,the integral imaging system can be divided into optical integral imaging system and integral imaging system with computer.The main idea of integral imaging technology is to acquire three-dimensional image array from objects by optical equipment such as microlens arrays or computer simulation of light path irradiation to obtain elemental image array,and then display the elemental image array to obtain the reconstructed three-dimensional object.At present,the existing methods of generating elemental image arrays at home and abroad mainly use lens array to collect objects and directly obtain elemental image array.However,this method is often affected by lens array technology,so it cannot collect larger objects and is difficult to be popularized.The camera array optical acquisition method needs to place and adjust a large number of cameras.In addition,using computer simulation of camera lens arrays and camera array optical path principle,the method of virtual collection of virtual objects is also applied to the acquisition of elemental image arrays.Although this approach avoids the limitation of optical devices during the process of collecting,it also has some disadvantages.It cannot collect and reconstruct real scene and the real objects.In this paper,a camera array acquisition method with computer-aided generation of elemental image array is adopted.Firstly,the spherical camera array is used to capture and collectreal objects and scenes,and then Matlab is used to program the two-dimensional images captured to generate elemental image array.However,as the camera array used in this paper are sparse camera array,the image super-resolution restoration method based on convolutional neural network is adopted to perform super-resolution reconstruction of the elemental image array,which solves the problem of fewer view images caused by sparse camera arrays.The integral imaging system in this paper includes two processes: acquisition process and display process.(1)Acquisition process: the acquisition system in this paper is composed of a camera array distributed on the surface of the sphere and a number of light sources to take photos of real objects and obtain multi-view images.Then,according to the mapping relationship between the multi-view images and the elemental image array,the elemental image array is acquired by programming.In order to achieve the effect of obtaining elemental image array generated by a large number of cameras with only a few cameras,this paper introduces the method of super resolution restoration,and the elemental image array generated by sparse camera is restored with super resolution to obtain a high-resolution elemental image array.(2)Display process: the display platform in this thesis is set up using the optical display platform combining liquid crystal displays and micro lens array.The elemental image array generated by image acquisition process is input to the liquid crystal display,and the display size of the elemental image array is adjusted in accordance with the micro lens array size.Finally the stereoscopic display effect can be watched from different views.The innovation of this thesis lies in the realization of spherical camera array to collect real objects,obtain elemental image array,and display on the display platform.The display results show that the proposed algorithm is feasible,and can achieve real scene acquisition and three-dimensional display of integral imaging in a real sense.