Research On Improving The Quality Of Integral Imaging Display

80% of the information humans obtain from the external world comes through the human eyes.Due to the binocular effect,the visual information obtained by the human eyes is a three-dimensional stereoscopic image with depth information.In stereoscopic display technology,naked eye 3D technology,especially integral imaging technology,is a popular research topic.It has good development prospects due to its advantages such as no need for viewing devices and no visual fatigue.However,due to the current limitations in computer processing power and optical device manufacturing processes,there is still room for improvement in integral imaging technology in areas such as image capture,computer reconstruction,and optical display.In terms of capturing and recording three-dimensional scenes,integral imaging technology uses camera arrays to record scene information,but it faces challenges such as low camera capture efficiency and complex camera array structures.In terms of imaging,resolution of reconstructed images is still relatively low and needs further improvement to meet visual demands.In traditional integral imaging methods,a large number of cameras are required for rendering.In order to improve the efficiency of camera acquisition,reduce the complexity of camera array,and not reduce image quality,this paper proposes a virtual stereo image synthesis method based on deep learning.This method extracts the feature values of the stereo image captured by the camera through a neural network,thereby generating a virtual stereo image without holes to increase the number of stereo images.Then,the original image and the synthesized virtual image are concatenated into a three-dimensional metaimage array according to their positions,and integral imaging reconstruction is carried out through a computer.The experimental results show that this method not only improves the acquisition efficiency of the camera,but also improves the display resolution.To further improve the display resolution,this paper proposes an integral imaging method based on image super-resolution and viewpoint synthesis.Using the image super-resolution method to super-resolution elemental images to obtain a high-resolution array of elemental images,and finally reconstruct high-quality integral imaging images.To verify the feasibility of the method proposed in this article,a camera array and scene model were constructed using Blender software,and the target object was captured to obtain a stereo image.The captured stereo image was used as a reference to synthesize a virtual stereo image.Using the method described in this article,super-resolution reconstruction of elemental images is performed,and then the images are concatenated into a high-resolution array of elemental images,ultimately achieving the reconstruction of the 3D scene.Experimental results show that this method can improve the display resolution of integral imaging.