Table 5 Quantitative comparison of different methods on the Chikusei dataset.

From: Cross-range self-attention single hyperspectral image super-resolution method based on U-Net architecture

Models

d

HR_image_size

\(\text {PSNR}\uparrow\)

\(\text {SSIM}\uparrow\)

\(\text {SAM}\downarrow\)

\(\text {ERGAS}\downarrow\)

\(\text {CC}\uparrow\)

Bicubic

4

256 × 256

37.4098

0.9021

13.6475

8.2598

0.9293

GDRRN

4

256 × 256

37.9635

0.9123

13.0025

7.8631

0.9358

Deep_hs_Prior

4

256 × 256

38.1485

0.9233

12.8954

7.3684

0.9385

SSPSR

4

256 × 256

38.3621

0.9232

12.6759

7.4123

0.9398

PDE-Net

4

256 × 256

37.6435

0.9242

13.2578

8.0265

0.9334

PUDPN

4

256 × 256

38.4805

0.9246

12.0133

7.4832

0.9426

CST

4

256 × 256

38.9124

0.9306

11.6547

7.1465

0.9480

Cs_Unet (Ours)

4

256 × 256

38.9427

0.9332

11.5610

7.1167

0.9491

Back to article page