Table 2.

Comparison of proposed method with other methods in DRIVE database.

Type	Method	Sensitivity	Specificity	Accuracy
	Human observer	0.7760	0.9724	0.9472
Unsupervised methods	Khawaja [1]	0.8043	0.9730	0.9553
	Mendonca [12]	0.7344	0.9764	0.9452
	Espona [13]	0.7436	0.9615	0.9352
	Vlachos [14]	0.7468	0.9551	0.9285
	Miri [10]	–	–	0.9458
	Wang [11]	–	–	0.9461
	Azzopardi [40]	0.7655	0.9704	0.9442
	Wang [41]	0.7527	–	0.9457
	Khawaja [39]	0.7907	0.9790	0.9603
	Roychowdhury [42]	0.7250	0.9830	0.9520
Supervised Methods	Liskowsk [43]	0.7763	0.9768	0.9495
	You [17]	0.7410	0.9751	0.9434
	Fraz [19]	0.7406	0.9807	0.9480
	Fu [24]	0.7294	–	0.9470
	Li [44]	0.7569	0.9816	0.9527
	Maninis [45]	0.8261	0.9115	0.9541
	Chen [46]	0.7426	0.9735	0.9453
	Orlando [20]	0.7079	0.9802	–
	Dasgupta [4]	0.7691	0.9801	0.9533
	Hu [47]	0.7772	0.9793	0.9533
	Na [48]	0.7680	0.9700	0.9540
	Proposed method	0.7762	0.9835	0.9694