@article{MB51F83BB,
title = "Image Stripe Noise Removal Method Using Haar Wavelet-Based MPRNet Architecture",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2024",
issn = "1226-4717",
doi = "10.7840/kics.2024.49.1.23",
author = "Byeongho Moon, Eunjae Ha, HaeMoon Kim, Byungin Choi",
keywords = "Stripe Noise, Noise Removal, Haar Wavelet Transform, Image Restoration, Deep Learning",
abstract = "Infrared imaging systems are widely used in surveillance and various other applications due to their ability to measure thermal information, which provides insights not visible with a visible light camera. However, due to the characteristics of the sensor, specific directional stripe noise can occur. Therefore, it is necessary to remove this type of noise. In this paper, we propose an effective method for removing this specific stripe noise by using a network structure that applies Haar Wavelet transformation and incorporates the MPRNet structure, known for its effectiveness in deep learning-based image restoration. The main contributions of our proposed method are as follows: First, We utilize Wavelet transformation taking into account the characteristics of stripe pattern noise. Second, We introduce an outstanding single-stage deep learning network for image restoration. Finally, We employ a loss function that considers the properties of Wavelet-transformed images.
The proposed network demonstrates superior performance in removing stripe noise from images compared to existing methods both qualitatively and quantitatively."
}