Table 2.

Average Recall Difference for the algorithms evaluated with the framework presented in Section 3 (bold values correspond to the algorithm that has the best relative performance in LWIR for the tested transformation).

ARD	BRISK	ORB	SIFT	BRIEF	FREAK	SURF
Blur	0.0442	−0.1323	0.1064	0.1149	0.0904	0.1425
Rotation	0.0450	−0.0762	0.0726	0.0109	0.0584	0.0013
Noise	−0.0427	−0.2921	−0.0764	−0.1266	−0.1273	−0.1106
Scale	0.0598	−0.0250	0.1564	0.0853	0.1271	0.1126