Light Field Angular Super-Resolution using Convolutional Neural Network with Residual Network
Light-field images are used in many areas because of its advantage that being able to film multiple images at once, but they are limited in their actual use due to their low resolution. To overcome this drawback, Light-field super-resolution studies are actively being conducted to expand the resolution of light-field images. Light-field super-resolution are divided into two categories. The first is to increase the spatial resolution and the other one is to increase angular resolution. In this paper, we propose an angular super-resolution algorithm based convolution neural network (CNN). Experimental results show that the proposed algorithm produces higher PSNR and SSIM values than existing methods.
Paper
Full text
Light Field Angular Super-Resolution using Convolutional Neural Network with Residual Network
Semantic Scholar · Computer Science · 2019
Abstract
Light-field images are used in many areas because of its advantage that being able to film multiple images at once, but they are limited in their actual use due to their low resolution. To overcome this drawback, Light-field super-resolution studies are actively being conducted to expand the resolution of light-field images. Light-field super-resolution are divided into two categories. The first is to increase the spatial resolution and the other one is to increase angular resolution. In this paper, we propose an angular super-resolution algorithm based convolution neural network (CNN). Experimental results show that the proposed algorithm produces higher PSNR and SSIM values than existing methods.