A Novel Technique Based on Visual Words Fusion Analysis of Sparse Features for Effective Content-Based Image Retrieval

Table 2

Performance analysis of the proposed technique based on visual words fusion on the Corel-A image collection which is reported using dictionary size of 800 visual words and features percentage of 50% per image (bold values indicate the best performance).

Semantic category	Proposed technique based on the visual words fusion	SIFT-LBP [41]	LGH-BoVW [36]	Color SIFT-EODH [42]	Poursistani et al. [43]	Yildizer et al. [44]
Africa	73.20	57.0	73.03	74.60	70.24	50.00
Beach	75.00	58.0	74.58	37.80	44.44	70.00
Buildings	80.30	43.0	80.24	53.90	70.80	20.00
Buses	95.50	93.0	95.84	96.70	76.30	80.00
Dinosaurs	100	98.0	97.95	99.00	100	90.60
Elephants	87.40	58.0	87.64	66.00	63.80	60.00
Flowers	98.30	83.0	85.13	92.00	92.40	100.00
Horses	97.10	68.0	86.29	87.00	94.70	80.00
Mountains	83.80	46.0	82.43	58.50	56.20	50.00
Food	82.40	53.0	78.96	62.20	74.50	20.00
MAP	87.30	65.7	84.21	72.77	74.34	62.00