Robust Retinal Blood Vessel Segmentation Based on Reinforcement Local Descriptions

Table 3

Segmentation results of different methods on DRIVE database.

Type	Methods	Year	SEN	SPE	ACC
Supervised methods	2nd human observer	—	0.7796	0.9717	0.9470
	Staal et al. [15]	2004	N.A	N.A	0.9442
	Soares et al. [16]	2006	N.A	N.A	0.9466
	Ricci and Perfetti [17]	2007	N.A	N.A	0.9563
	Welikala et al. [24]	2014	0.737	0.9521	0.942
	Nguyen et al. [25]	2013	0.7321	0.9487	0.9407
	Lupascu et al. [19]	2010	0.7200	N.A	0.9597
	Marín et al. [20]	2011	0.7067	0.9801	0.9452
	You et al. [8]	2011	0.7410	0.9751	0.9434
	Fraz et al. [21]	2012	0.7406	0.9807	0.9480
	Li et al. [22]	2016	0.7569	0.9816	0.9527
	Proposed method	2016	0.7680	0.9827	0.9630
Unsupervised methods	Mendonça and Campilho [34]	2006	0.7344	0.9764	0.9452
	Alonso-Montes et al. [28]	2008	N.A	N.A	0.9185
	Lam et al. [9]	2010	N.A	N.A	0.9472
	Delibasis et al. [35]	2010	N.A	N.A	0.9407
	Qian Zhao et al. [13]	2014	0.7354	0.9789	0.9477