Skip to main content
. 2022 Sep 30;12:16459. doi: 10.1038/s41598-022-20816-1

Table 1.

Quantitative comparison of the tested methods.

Method Mean RE (↓) Mean normalized RE (↓) Mean SSIM (↑)
No transformation 41.77 1 0.431
Manual annotation
SH 8.44 0.267 0.513
MLS 6.20 0.257 0.531
BEBLID w/o 256 512 w/o 256 512 w/o 256 512
A-KAZE
SH 10.87 10.77 10.88 0.346 0.350 0.355 0.526 0.528 0.528
MLS 25.13 44.89 34.92 0.690 1.104 0.851 0.519 0.489 0.504
SH + HMA 10.88 10.82 10.89 0.346 0.355 0.357 0.518 0.521 0.521
ORB 1000
SH 11.54 12.20 12.16 0.374 0.395 0.390 0.520 0.523 0.523
MLS 57.17 46.01 38.98 1.465 1.214 1.023 0.480 0.494 0.503
SH + HMA 11.54 12.21 12.26 0.374 0.395 0.391 0.517 0.520 0.519
BRISK
SH 11.35 11.52 11.51 0.366 0.367 0.361 0.529 0.527 0.527
MLS 51.02 38.41 30.98 1.240 1.006 0.820 0.487 0.494 0.511
SH + HMA 11.36 12.68 11.52 0.365 0.380 0.361 0.521 0.520 0.520
MG-DHNN25 20.18 0.549 0.515
MG-DHNN fine-tuned 13.17 0.408 0.514

First and second-best solution are printed in bold and italic respectively.