Low-level structure feature extraction for image processing via stacked sparse denoising autoencoder

•We use deep learning for image processing by extracting image feature.•The extraction has good performance in noisy or low-light circumstance.•We optimize TV and L0 smoothing filter by the proposed feature extraction.•The features can extract image structure features regardless the inputs directly....

Full description

Saved in:
Bibliographic Details
Published in:Neurocomputing (Amsterdam) Vol. 243; pp. 12 - 20
Main Authors: Fan, Zunlin, Bi, Duyan, He, Linyuan, Shiping, Ma, Gao, Shan, Li, Cheng
Format: Journal Article
Language:English
Published: Elsevier B.V 21.06.2017
Subjects:
Deep learning
Image enhancement
Image processing
Pattern recognition
Spatial filter
Deep learning
Pattern recognition
Image processing
Spatial filter
Image enhancement
ISSN:0925-2312, 1872-8286
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•We use deep learning for image processing by extracting image feature.•The extraction has good performance in noisy or low-light circumstance.•We optimize TV and L0 smoothing filter by the proposed feature extraction.•The features can extract image structure features regardless the inputs directly. In this paper, we propose a novel low-level structure feature extraction for image processing based on deep neural network, stacked sparse denoising autoencoder (SSDA). The current image processing methods via deep learning are directly building and learning the end-to-end mappings between the input/output. Instead, we advocate the analysis of the first layer learning features from input data. With the learned low-level structure features, we improve two edge-preserving filters that are key to image processing tasks such as denoising, High Dynamic Range (HDR) compression and details enhancement. Due to the validity and superiority of the proposed feature extraction, the results computed by the two improved filters do not suffer from the drawbacks including halos, edge blurring, noise amplification and over-enhancement. More importantly, we demonstrate that the features trained from natural images are not specific and can extract the structure features of infrared images. Hence, it is feasible to handle tasks by using the trained features directly.
ISSN:0925-2312
1872-8286
DOI:10.1016/j.neucom.2017.02.066