Depth-Normal Fusion on DiLiGenT-MV first view direction
0.353Bear RMSE (mm)PTGV
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PTGValpha=1, beta=1, lambda0=0.001, lambda1=0.001, type=perspective-aware TGV fusion2026.02 | 0.353 | 0.129 | 1.056 | 0.279 | 0.653 | 0.153 | 0.61 | 0.177 | 1.14 | 0.238 | 0.762 | 0.195 | |
| Nehabalpha=0.1, extension=handle missing depth data2026.02 | 0.392 | 0.53 | 2.731 | 0.672 | 0.596 | 0.572 | 0.467 | 0.566 | 2.313 | 0.635 | 1.3 | 0.586 | |
| BiNIlambda1=10^72026.02 | 0.463 | 0.398 | 1.847 | 0.463 | 0.663 | 0.417 | 0.53 | 0.416 | 1.068 | 0.451 | 0.914 | 0.429 | |
| PGalpha=1, beta=1, type=perspective-aware gradient-based fusion2026.02 | 0.634 | 0.371 | 1.753 | 0.446 | 1.229 | 0.412 | 0.781 | 0.373 | 1.705 | 0.414 | 1.22 | 0.403 | |
| Naivealpha=1, beta=1, approach=perspective-aware fusion using reprojection to orthographic2026.02 | 0.728 | 0.259 | 1.678 | 0.342 | 1.309 | 0.307 | 1.526 | 0.279 | 2.002 | 0.336 | 1.449 | 0.305 | |
| Orthoalpha=1, beta=1, approach=pure orthographic fusion applied directly to perspective depth maps2026.02 | 1.506 | 0.431 | 2.245 | 0.5 | 1.656 | 0.468 | 2.746 | 0.445 | 2.616 | 0.489 | 2.154 | 0.467 |