Transferred Multi-Perception Attention Networks for Remote Sensing Image Supe...

杨利霞

Transferred Multi-Perception Attention Networks for

1 z5 B9 i% B6 U2 E! P: y

Remote Sensing Image Super-Resolution

0 q; E5 k6 p- ?) d! N
5 W) P3 A% @5 O, `" n

% M6 H" n' q4 r  f- {2 fImage super-resolution (SR) reconstruction plays a key role in coping with the increasing
3 g+ }% ~+ ^: Kdemand on remote sensing imaging applications with high spatial resolution requirements. Though
many SR methods have been proposed over the last few years, further research is needed to improve
% R3 c! b7 e8 |! M. K( e5 E8 wSR processes with regard to the complex spatial distribution of the remote sensing images and the
: Z$ t( i$ g, ^diverse spatial scales of ground objects. In this paper, a novel multi-perception attention network
8 s( ?0 u2 M3 q0 y8 V6 I4 [4 s(MPSR) is developed with performance exceeding those of many existing state-of-the-art models.
By incorporating the proposed enhanced residual block (ERB) and residual channel attention group
(RCAG), MPSR can super-resolve low-resolution remote sensing images via multi-perception learning
and multi-level information adaptive weighted fusion. Moreover, a pre-train and transfer learning
# A- X1 d5 g. b, W. Q& Pstrategy is introduced, which improved the SR performance and stabilized the training procedure.
# d$ ?% w: |# y6 ^+ M1 Z# L; UExperimental comparisons are conducted using 13 state-of-the-art methods over a remote sensing
6 z( X/ ~+ k/ o% @dataset and benchmark natural image sets. The proposed model proved its excellence in both objective
criterion and subjective perspective.