Table 4 Recognition rates and and per-object detection times (where available) for the UWA scenes
From: Local shape feature fusion for improved matching, pose estimation and 3D object recognition
Method | Recognition rate | Timing (s) |
|---|---|---|
External systems | ||
Spin Images | 0.878 | \(\sim\)7200 |
Tensor Matching | 0.966 | \(\sim\)90 |
PPF, \(\tau _d=0.025\) | 0.970 | 85 |
PPF, \(\tau _d=0.04\) | 0.892 | 1.97 |
EM | 0.975 | – |
RoPS based | 0.989 | – |
RANSAC, single feature | ||
ECSAD | 0.947 | 1.139 |
NDHist | 0.968 | 1.155 |
RoPS | 0.846 | 1.254 |
SHOT | 0.809 | 1.221 |
SI | 0.803 | 0.998 |
RANSAC, ternary feature | ||
ECSAD+NDHist+RoPS | 0.979 | 1.253 |
ECSAD+NDHist+SHOT | 0.979 | 1.192 |
ECSAD+NDHist+SI | 0.968 | 1.116 |
ECSAD+RoPS+SHOT | 0.947 | 1.187 |
ECSAD+RoPS+SI | 0.957 | 1.095 |
ECSAD+SHOT+SI | 0.957 | 1.103 |
NDHist+RoPS+SHOT | 0.952 | 1.182 |
NDHist+RoPS+SI | 0.957 | 1.102 |
NDHist+SHOT+SI | 0.984 | 1.112 |
RoPS+SHOT+SI | 0.931 | 1.227 |