Image Retrieval Using Low Level and Local Features Contents: A Comprehensive Review

Table 1

Summary of color feature-based image retrieval techniques.

Year	Method	Similarity measure	Dataset	Performance measure (%)
2015	Dot-diffused BTC [10]	Modified Canberra	Corel-1000	Accuracy: 77.16
			Brodatz-1856	Accuracy: 81.19
			VisTex-640	Accuracy: 92.09
			STex	Accuracy: 44.79
			ALOT	Accuracy: 48.64
			OutexTC00013	Accuracy: 66.82
			KTHTIPS	Accuracy: 64.81
2016	Feature vector generation using Niblack binarizaion, classification using artificial neural network [18]	City block distance	Wang	Precision: 83.8 Recall: 83.7
			OT scene	Precision: 66.5
				Recall: 66.3
2018	Color histogram using quantized HSV color space [19]	Improved dominance granule structure similarity method	COIL-20	Precision: 48.18
				Recall: 83.87
			Corel-1000	Precision: 68.3
				Recall: 37.9
2015	Error diffusion BTC, “color histogram feature” (CHF), and “bit pattern histogram feature” [11]	Modified Canberra	Corel-1000	Precision: 79.7
			Corel-10000	Precision: 79.8
2018	BTC based on binary ant colony optimization [14]	Modified Canberra	Corel-1000	Precision: 80.565
			Corel-10000	Precision: 65
2008	Quantized HSV color space histogram and dominant color descriptor [5]	L1 distance	Three categories from Corel-1000	Precision: 89.64
				Recall: 76.47