Table 3 Average FSIM of different algorithms on PatternNet, AID, WHU-RS19,NWPURESISC45,UCMERCED.

From: Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer

Dataset

Scale

Bicubic

SRGAN

ESRGAN

SRTransGAN

Proposed algorithm

PatternNet

×2

0. 861

0.988

0. 994

0. 997

0. 997

×3

0. 850

0.989

0. 990

0. 993

0. 996

×4

0. 834

0.976

0. 981

0. 986

0. 993

AID

×2

0. 824

0.989

0. 904

0. 908

0. 986

×3

0. 814

0.896

0. 900

0. 903

0. 903

×4

0. 798

0.887

0. 892

0. 897

0. 899

WHU-RS19

×2

0. 832

0.907

0. 910

0. 914

0. 915

×3

0. 822

0.902

0. 906

0. 909

0. 912

×4

0. 806

0.894

0. 898

0. 903

0. 909

NWPURESISC45

×2

0.841

0.913

0.920

0.926

0.947

×3

0.830

0.901

0.914

0.921

0.912

×4

0.809

0.895

0.906

0.917

0.929

UCMERCED

×2

0.827

0.903

0.917

0.935

0.974

×3

0.804

0.891

0.911

0.929

0.938

×4

0.796

0.874

0.904

0.923

0.942

Back to article page