Table 5 Performance evaluation on the texture feature-based variants of our algorithm with features detected on the original reflectance images and on atlases of rectified reflectance images
From: Robust surface registration using N-points approximate congruent sets
|
| |||||
|---|---|---|---|---|---|---|
Δ | S r (%) | t (s) | Δ | S r (%) | t (s) | |
Uniform | 0.0727 | 100 | 32.2 | ★ | 0 | 53.1 |
SURF + uniform | 0.0349 | 100 | 26.2 | 0.2632 | 60 | 11.0 |
matchSURF(k = 500) + uniform | 0.0353 | 100 | 15.9 | 0.2555 | 55 | 4.6 |
matchSURF(k = 200) + uniform | 0.0568 | 100 | 7.7 | 0.2532 | 40 | 4.6 |
matchSURF(k = 100) + uniform | 0.0952 | 100 | 4.7 | 0.3667 | 50 | 4.4 |
matchSURF(k = 50) + uniform | 0.1592 | 90 | 4.5 | 0.3201 | 25 | 3.1 |
Atlas: SURF + uniform | 0.0148 | 100 | 17.8 | 0.2018 | 100 | 8.7 |
Atlas: matchSURF(k = 500) + uniform | 0.0199 | 100 | 6.2 | 0.1560 | 85 | 6.1 |
Atlas: matchSURF(k = 200) + uniform | 0.0248 | 100 | 5.1 | 0.2299 | 95 | 4.3 |
Atlas: matchSURF(k = 100) + uniform | 0.0318 | 100 | 4.9 | 0.3095 | 75 | 4.0 |
Atlas: matchSURF(k = 50) + uniform | 0.0367 | 100 | 2.5 | 0.3354 | 60 | 3.3 |
SIFT + uniform | 0.0433 | 100 | 34.4 | 0.2351 | 30 | 11.5 |
matchSIFT(k = 500) + uniform | 0.0548 | 100 | 15.3 | 0.2222 | 10 | 10.4 |
matchSIFT(k = 200) + uniform | 0.0849 | 100 | 9.5 | 0.2801 | 10 | 3.9 |
matchSIFT(k = 100) + uniform | 0.1674 | 100 | 5.9 | ★ | 0 | 3.4 |
matchSIFT(k = 50) + uniform | 0.2842 | 85 | 5.5 | ★ | 0 | 1.5 |
Atlas: SIFT + uniform | 0.0287 | 100 | 28.7 | 0.0494 | 30 | 10.8 |
Atlas: matchSIFT(k = 500) + uniform | 0.0321 | 100 | 29.4 | 0.1584 | 25 | 4.3 |
Atlas: matchSIFT(k = 200) + uniform | 0.0484 | 100 | 10.2 | 0.3768 | 15 | 5.3 |
Atlas: matchSIFT(k = 100) + uniform | 0.1716 | 100 | 5.2 | 0.5138 | 10 | 4.4 |
Atlas: matchSIFT(k = 50) + uniform | 0.3017 | 90 | 5.6 | ★ | 0 | 1.9 |