Entropy-Based and Weighted Selective SIFT Clustering as an Energy Aware Framework for Supervised Visual Recognition of Man-Made Structures

Table 1

Summary of results for the ZuBuD database.

Method	Keypoints in reference images	Keypoints in query image	Ratio of selected keypoints to total keypoints	Recognition rate	Ratio of computational energy costs
Work in [2]	All in filtered region	All in filtered region	—	94.80%	n/a
Work in [1]	All in filtered region	All in filtered region	—	96.50%	0.52%
Relevant keypoints	20	20	2.7%	91.30%	0.07%
	30	30	4.1%	94.80%	0.16%
	40	40	5.4%	95.70%	0.29%
	50	50	6.8%	96.50%	0.46%
	100	75	10.1%	98.30%	1.37%
	100	115	15.5%	99.10%	2.10%