Inception residual attention network for remote sensing image super-resolution

杨利霞

Inception residual attention network for remote sensingimage super-resolution

/ o0 O" ?4 V! F$ Z
# Y. N& F$ t( H; d, L: J# Q* yHow to enhance the spatial resolution for a remote sensing image is
an important issue that we face. Many image super-resolution (SR)
, M+ L7 Z$ |& m' S& D& j2 [1 a: otechniques have been proposed for this purpose and deep con
volutional neural network (CNN) is the most effective approach in
recent years. However, we observe that most CNN-based SR meth
ods treat low-frequency areas and high-frequency areas equally,
hence hindering the recovery of high-frequency information. In this
paper, we propose a network named inception residual attention
network (IRAN) to address this problem. Specifically, we propose
a spatial attention module to make the network adaptively learn
$ G2 y6 ?* ?) P6 }% t. E: \the importance of different spatial areas, so as to pay more atten
: A+ z3 N: w  F  g8 Otion to the areas with high-frequency information. Furthermore, we
( k& y' K6 _+ c$ M/ z6 z% Ypresent an inception module to fuse local multilevel features, so as
' w; d& m8 |8 g0 zto provide richer information for reconstructing detailed textures. In
order to evaluate the effectiveness of the proposed method, a large
number of experiments are performed on UCMerced-LandUse data
6 V) v% ]# b3 B$ f6 wset and the results show that the proposed method is superior to
* B& M- w% g9 zthe current state-of-the-art methods in both visual effects and
1 ]. T! h2 B- o6 M8 V* {7 f& n' ?0 Eobjective indicators.

- p# S  H) U1 x7 I  Y


+ o7 Y' n3 d0 H8 E7 N4 |9 y