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

杨利霞

Transferred Multi-Perception Attention Networks for

Remote Sensing Image Super-Resolution

7 D% W1 z2 J1 M2 Z/ ?

$ r$ H3 E9 I, G) G. \Image super-resolution (SR) reconstruction plays a key role in coping with the increasing
8 ^$ C: R% h, }3 t3 D3 \; a# Edemand 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
$ b' T+ L. q' ~/ qSR processes with regard to the complex spatial distribution of the remote sensing images and the
' X5 c" ^9 |9 ~( l& O1 u/ K1 q  ^diverse spatial scales of ground objects. In this paper, a novel multi-perception attention network
, ?4 y% j7 \$ E5 u. v5 J$ I(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
) N4 P5 X8 B& z( B0 Y5 F7 R) Gand multi-level information adaptive weighted fusion. Moreover, a pre-train and transfer learning
; z  P9 n7 F+ V& k& c5 Vstrategy is introduced, which improved the SR performance and stabilized the training procedure.
Experimental comparisons are conducted using 13 state-of-the-art methods over a remote sensing
dataset and benchmark natural image sets. The proposed model proved its excellence in both objective
* e. O5 t5 P4 [7 K0 C& D1 bcriterion and subjective perspective.

$ E/ e5 e; K- W* c3 _# B