Abstract:Remote sensing images play an important role in many practical applications, however, due to the physical limitations of remote sensing devices, it is difficult to obtain images at an expecting high resolution level. Acquiring high-resolution(HR) images from the original low-resolution(LR) ones with super-resolution(SR) methods has always been an attractive proposition in embedded systems including various kinds of tablet PC and smart phone. SR methods based on sparse representation have been successfully used in processing remote sensing images, however, they have two major problems in common. First, they use only one type of image features to represent the low resolution(LR) images. However, one single type of features cannot accurately represent an image due to the diverse structures of the image, as a result, artifacts would be produced simultaneously. Second, many dictionary learning methods try to build a universal dictionary with only one single type of features. However, apparently, a dictionary with a single type of features is not enough to capture the different structures of a remote sensing image, without any doubt, the resultant image would turn out to be a poor one. To overcome the problems above, we propose a new framework for remote sensing image super resolution: sparse representation-based SR method by processing dictionaries with multi-type features. First, in order to represent the remote sensing image more accurately, different types of features are extracted from images. Second, to achieve a better performance, various dictionaries with multi-type features are learned to capture the essential structures of the image. Then, it's proposed to adaptively control the weights of the high resolution(HR) patches obtained by different dictionaries. Numerous experiments validate that this proposed framework brings better results in terms of both objective quantitation and visual perception than other compared algorithms. (C) 2015 Elsevier B.V. All rights reserved.

Single Image Super-Resolution Using Incoherent Sub-Dictionaries Learning

Single-Image Super-Resolution by Subdictionary Coding and Kernel Regression

A Sparse Representation Based Super-Resolution Image Reconstruction Scheme Utilizing Dual Dictionaries

Double Sparse Dictionary Learning for Image Super Resolution

Single Image Super-Resolution by Multi-Dictionary Learning

Single Image Super-Resolution Via Residual Dictionary Learning

Single Image Super-Resolution Based on Deep Learning Features and Dictionary Model.

Image Super-Resolution Using Multi-dictionary Sparse Representation

Super-resolution Reconstruction of Sparse Dictionaries Based on Integrated Feature.

Single image super resolution based on adaptive multi-dictionary learning

Remote Sensing Images Super-resolution Based on Sparse Dictionaries and Residual Dictionaries.

Single Image Super Resolution With High Resolution Dictionary

Learning Shared and Cluster-Specific Dictionaries for Single Image Super-Resolution

Single Image Super Resolution Based on Sparse Domain Selection

A New Framework for Remote Sensing Image Super-Resolution: Sparse Representation-Based Method by Processing Dictionaries with Multi-Type Features

Image Super-Resolution Via Dual-Dictionary Learning And Sparse Representation

Alternatively Constrained Dictionary Learning for Image Superresolution

Single Image Super-Resolution Via Projective Dictionary Learning with Anchored Neighborhood Regression

A Sparse Representation Based Approach for Super-Resolution Reconstruction over Geometric Dictionaries and Consistency of Gradients

Image Super-Resolution Based on Sparse Representation via Direction and Edge Dictionaries