Enhanced CNN for image denoising

Owing to the flexible architectures of deep convolutional neural networks (CNNs) are successfully used for image denoising. However, they suffer from the following drawbacks: (i) deep network architecture is very difficult to train. (ii) Deeper networks face the challenge of performance saturation....

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	CAAI Transactions on Intelligence Technology Jg. 4; H. 1; S. 17 - 23
Hauptverfasser:	Tian, Chunwei, Xu, Yong, Fei, Lunke, Wang, Junqian, Wen, Jie, Luo, Nan
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	Beijing The Institution of Engineering and Technology 01.03.2019 John Wiley & Sons, Inc Wiley
Schlagworte:	Artificial neural networks authors B6135 Optical, image and video signal processing batch normalisation techniques C5260B Computer vision and image processing techniques C5290 Neural computing techniques convolution convolutional neural denoising network deep convolutional neural networks deep network architecture Deeper networks dilated convolutions enhanced CNN flexible architectures image denoising Image enhancement image representation image restoration image restoration CNN learning (artificial intelligence) neural nets Noise Noise reduction performance saturation Research Article residual learning training difficulties dilated convolutions residual learning enhanced CNN convolutional neural denoising network batch normalisation techniques deep network architecture image denoising flexible architectures image restoration Deeper networks image representation training difficulties convolution performance saturation deep convolutional neural networks learning (artificial intelligence) image restoration CNN neural nets authors
ISSN:	2468-2322, 2468-6557, 2468-2322
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Beschreibung
Zusammenfassung:	Owing to the flexible architectures of deep convolutional neural networks (CNNs) are successfully used for image denoising. However, they suffer from the following drawbacks: (i) deep network architecture is very difficult to train. (ii) Deeper networks face the challenge of performance saturation. In this study, the authors propose a novel method called enhanced convolutional neural denoising network (ECNDNet). Specifically, they use residual learning and batch normalisation techniques to address the problem of training difficulties and accelerate the convergence of the network. In addition, dilated convolutions are used in the proposed network to enlarge the context information and reduce the computational cost. Extensive experiments demonstrate that the ECNDNet outperforms the state-of-the-art methods for image denoising.
Bibliographie:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14
ISSN:	2468-2322 2468-6557 2468-2322
DOI:	10.1049/trit.2018.1054