Hierarchical level features based trainable segmentation for electron microscopy images

Table 3 Experimental results of several methods on the ISBI2012’s data set

Method	Rand error	Warping error	Pixel error	Classifier input
Second human observer	0.026546995	0.000344086	0.066553289	N/A
Simple thresholding	0.449664478	0.017141342	0.225194944	N/A
Laptev’s method	--	0.00062	0.079264809	1878 dimension features
Burget’s method	0.139038440	0.002641296	0.102285508	Local-level and Segment-level features
HLFs-RF	0.106308715	0.001200104	0.079132453	72 dimension features

“--”means unknown, [Laptev et al., MICCAI’2012] do not indicate the Rand error. N/A means not available.

ISSN: 1475-925X