Single Infrared Image Stripe Noise Removal Using Deep Convolutional Networks

In this paper, we present a deep learning method for single infrared image stripe noise removal. Our method is denoted as a deep convolutional neural network (CNN) that takes the noisy image as the input and outputs the clean image. The deep CNN consists of two components: 1) image denoising, substa...

Full description

Saved in:
Bibliographic Details
Published in:IEEE photonics journal 2017-08, Vol.9 (4), p.1-13
Main Authors: Kuang, Xiaodong, Sui, Xiubao, Chen, Qian, Gu, Guohua
Format: Article
Language:English
Subjects:
Convolution
deep convolutional neural networks
Histograms
Image denoising
Image resolution
infrared image
Kernel
Neural networks
Stripe noise removal
Training
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we present a deep learning method for single infrared image stripe noise removal. Our method is denoted as a deep convolutional neural network (CNN) that takes the noisy image as the input and outputs the clean image. The deep CNN consists of two components: 1) image denoising, substantially removing the stripe noise but losing details, 2) image denoising and super resolution, completely eliminating the residual stripe noise and restore details. Our deep CNN exhibits excellent image denoising and detail preserving performance. Meanwhile it achieves fast speed for real-time image processing. Experiments study the relationship between model parameter settings and model performance.
ISSN:1943-0655
1943-0647
DOI:10.1109/JPHOT.2017.2717948