Pixel-Level Kernel Estimation for Blind Super-Resolution

Lew, Jaihyun; Kim, Euiyeon; Heo, Jae-Pil

doi:10.1109/access.2021.3128272

Cited by 3 publications

(2 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…J. Lew et al, [7] calculations unavoidably become substandard in taking care of situations where a picture is corrupted non-consistently. A separation and-vanquish type of approach, isolating a picture into a few patches for individual portion assessment and SR can be a basic answer for this.…”

Section: Literature Surveymentioning

confidence: 99%

An Extensive Comparative Analysis on Various Efficient Techniques for Image Super-Resolution

Bhujade¹,

Asthana²

2022

IJETAE

View full text Add to dashboard Cite

The image up scaling and super resolution gain a attention now a days. Create a high-resolution image with help of a low-resolution image uses in many of the application. The optimization based super-resolution methods is principally driven by the choice of the objective function. Recently artificial intelligence based machine and deep learning methods are also using for the image processing application. It makes easy and efficient way of image super resolution. Recent work has generally centered around limiting the mean squared reproduction blunder. The subsequent evaluations have high pinnacle signal-tocommotion proportions, however they are much of the time lacking high-recurrence subtleties and are perceptually subpar as in they neglect to match the constancy expected at the higher resolution. This paper presents the comparative analysis of the efficient techniques for image super-resolution. Keywords—Image, Super, Resolution, AI, Machine learning, Deep Learning, Perfromance.

show abstract

Section: Literature Surveymentioning

confidence: 99%

An Extensive Comparative Analysis on Various Efficient Techniques for Image Super-Resolution

Bhujade¹,

Asthana²

2022

IJETAE

View full text Add to dashboard Cite

show abstract

“…Knowledge distillation methods(Zhang et al 2021a;Wang et al 2021;Lee et al 2020;Gao et al 2019) have shown their efficiency on small SR networks where(Lee et al 2020) uses privileged information to boost the teacher network's performance. Meanwhile,(Lew, Kim, and Heo 2021;Gu et al 2019;Bell-Kligler, Shocher, and Irani 2019) aims to model the complex degradations explicitly. To tackle the ill-posed nature of SISR, several methods(Ledig et al 2017;Wang et al 2018;Zhang et al 2019) obtain enhanced visual quality by utilizing the adversarial loss and the perceptual loss(Johnson, Alahi, and Fei-Fei 2016).…”

mentioning

confidence: 99%

Organic, Inorganic, Hybrid, and Bio Materials Research at Sungkyunkwan University

Lee

2019

Advanced Materials

View full text Add to dashboard Cite

Recent deep-learning-based single image super-resolution (SISR) methods have shown impressive performance whereas typical methods train their networks by minimizing the pixelwise distance with respect to a given high-resolution (HR) image. However, despite the basic training scheme being the predominant choice, its use in the context of ill-posed inverse problems has not been thoroughly investigated. In this work, we aim to provide a better comprehension of the underlying constituent by decomposing target HR images into two subcomponents: (1) the optimal centroid which is the expectation over multiple potential HR images, and (2) the inherent noise defined as the residual between the HR image and the centroid. Our findings show that the current training scheme cannot capture the ill-posed nature of SISR and becomes vulnerable to the inherent noise term, especially during early training steps. To tackle this issue, we propose a novel optimization method that can effectively remove the inherent noise term in the early steps of vanilla training by estimating the optimal centroid and directly optimizing toward the estimation. Experimental results show that the proposed method can effectively enhance the stability of vanilla training, leading to overall performance gain. Codes are available at github.com/2minkyulee/ECO.

show abstract