Inception residual attention network for remote sensing image super-resolution

杨利霞

Inception residual attention network for remote sensingimage super-resolution

$ M  g% s0 r5 R& Q+ j% X! b  @

How to enhance the spatial resolution for a remote sensing image is
' F6 \6 X5 l- n$ j" s, P: Nan important issue that we face. Many image super-resolution (SR)
techniques have been proposed for this purpose and deep con
) S) p- }+ m& M/ hvolutional neural network (CNN) is the most effective approach in
7 ^9 f: t2 a. ]; X' p3 Lrecent years. However, we observe that most CNN-based SR meth
/ P! U5 s, D2 \* v/ W9 Gods treat low-frequency areas and high-frequency areas equally,
hence hindering the recovery of high-frequency information. In this
3 {: n; ?$ e& Lpaper, we propose a network named inception residual attention
9 V/ |7 W8 }" W5 H" ~0 ?, A! cnetwork (IRAN) to address this problem. Specifically, we propose
a spatial attention module to make the network adaptively learn
the importance of different spatial areas, so as to pay more atten
tion to the areas with high-frequency information. Furthermore, we
present an inception module to fuse local multilevel features, so as
0 l5 A0 t/ n8 e/ r7 v2 z7 mto provide richer information for reconstructing detailed textures. In
order to evaluate the effectiveness of the proposed method, a large
+ V6 ]0 o, Y- T: t% @number of experiments are performed on UCMerced-LandUse data
set and the results show that the proposed method is superior to
6 @9 t& w# ^) U% k9 C6 `; K4 n* vthe current state-of-the-art methods in both visual effects and
objective indicators.



; S+ A1 w; w, M8 T7 J( q