数学建模社区-数学中国

标题: Transferred Multi-Perception Attention Networks for Remote Sensing Image Supe... [打印本页]
作者: 杨利霞    时间: 2020-11-13 16:10
标题: Transferred Multi-Perception Attention Networks for Remote Sensing Image Supe...
Transferred Multi-Perception Attention Networks for

1 [# t! f. \' c4 H, B
Remote Sensing Image Super-Resolution
: H, v; T, v. l! x2 ?* L
2 @/ O9 D1 ]* G# ^. g, k1 b

  ]! d$ r) Y' U( I2 e6 \
  e5 n/ ^4 n- t2 C) d: u5 r/ e1 [8 hImage super-resolution (SR) reconstruction plays a key role in coping with the increasing9 S; F2 u) D3 @* t5 @$ r+ o) R
demand on remote sensing imaging applications with high spatial resolution requirements. Though4 u* C  P; B: Q/ }6 l
many SR methods have been proposed over the last few years, further research is needed to improve
) ^7 ~" |7 m7 S; v5 {SR processes with regard to the complex spatial distribution of the remote sensing images and the  }5 E, S3 b$ u& k
diverse spatial scales of ground objects. In this paper, a novel multi-perception attention network
- G6 m( v& l* Q  X" G6 ?(MPSR) is developed with performance exceeding those of many existing state-of-the-art models./ S, k9 M$ l+ {' ^6 P
By incorporating the proposed enhanced residual block (ERB) and residual channel attention group: q) ?3 g9 }" S8 @5 T( I. f. L  H
(RCAG), MPSR can super-resolve low-resolution remote sensing images via multi-perception learning
* f2 h. v$ y; H: @6 D0 i' band multi-level information adaptive weighted fusion. Moreover, a pre-train and transfer learning! C  {0 e' e* }
strategy is introduced, which improved the SR performance and stabilized the training procedure.
- Q2 {+ F! v, q2 X$ _6 V% fExperimental comparisons are conducted using 13 state-of-the-art methods over a remote sensing, S4 l7 S. b2 m2 S, ]; {: ]
dataset and benchmark natural image sets. The proposed model proved its excellence in both objective
4 y8 U6 n, R5 C& _criterion and subjective perspective.
: O3 J0 S- P0 n# h8 U. _- Y7 w: q9 F5 @1 R9 k. z" ?

& Q5 l% [% D$ u- H

Transferred Multi-Perception Attention Networks for.pdf

11.42 MB, 下载次数: 0, 下载积分: 体力 -2 点



欢迎光临 数学建模社区-数学中国 (http://www.madio.net/) Powered by Discuz! X2.5