Inception residual attention network for remote sensing image super-resolution

杨利霞

Inception residual attention network for remote sensingimage super-resolution

How to enhance the spatial resolution for a remote sensing image is
an important issue that we face. Many image super-resolution (SR)
; k- T- V3 k2 K' A& j" D7 F7 Dtechniques have been proposed for this purpose and deep con
volutional neural network (CNN) is the most effective approach in
  A4 [- H' n( n+ Orecent 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
0 ?% ]: ?- i: N7 m0 p0 Mnetwork (IRAN) to address this problem. Specifically, we propose
# \& [1 s. X2 na spatial attention module to make the network adaptively learn
. O, E- w/ |  f$ q6 Bthe importance of different spatial areas, so as to pay more atten
  r/ x! a$ I# t. Q7 {tion to the areas with high-frequency information. Furthermore, we
present an inception module to fuse local multilevel features, so as
to provide richer information for reconstructing detailed textures. In
% W- G2 v9 j& E! d% d7 j  Q% sorder to evaluate the effectiveness of the proposed method, a large
2 h) I) B, U( V5 S. O: A. X' Fnumber of experiments are performed on UCMerced-LandUse data
" a6 ^) e( H2 Qset and the results show that the proposed method is superior to
* i: `' z$ x, x: Y" Jthe current state-of-the-art methods in both visual effects and
objective indicators.
. e' D$ `; h( z( W: \; \


" i5 |1 W/ D' T& a
* y  K) N% B* @8 @& @. N1 y