Table 3.

Classification results of five image features using different classifiers in the 160 × 160 pixels sub-database of GasHisSDB [In (%)].

Freatures	Methods	Acc			Abnormal						Normal
					Pre	Rec	Spe	F1			Pre	Rec	Spe	F1
	LR	83.29			81.32	80.42	85.54	80.87			84.80	85.54	80.42	85.17
	kNN	85.52			82.95	84.35	86.43	83.64			87.58	86.43	84.35	87.01
	RF	85.99			81.65	87.83	84.55	84.63			89.88	84.55	87.83	87.13
Color histogram	Linear SVM	41.12			33.92	35.96	45.16	34.91			47.40	45.16	35.96	46.25
	Non-linear SVM	56.09			Null	0.00	100.00	0.00			56.09	100.00	0.00	71.87
	ANN	78.89			77.78	72.68	83.75	75.14			79.67	83.75	72.68	81.66
	LR	70.97			67.95	49.92	84.67	57.56			72.21	84.67	49.92	77.95
	kNN	77.10			70.30	72.60	80.03	71.43			81.78	80.03	72.60	80.90
	RF	79.13			72.17	76.60	80.78	74.32			84.14	80.78	76.60	82.42
	Linear SVM	42.34			40.50	98.67	5.68	57.43			86.74	5.68	98.67	10.66
Luminance histogram	Non-linear SVM	60.58			Null	0.00	100.00	0.00			60.58	100.00	0.00	75.45
	ANN	71.23			64.74	59.34	78.97	61.92			74.90	78.97	59.34	76.88
HOG	LR	60.46			48.96	7.20	95.11	12.56			61.16	95.11	7.20	74.45
	kNN	61.42			51.31	41.65	74.28	45.98			66.17	74.28	41.65	69.99
	Naive Bayesian	54.43			45.11	71.84	43.11	55.42			70.17	43.11	71.84	53.40
	RF	60.85			50.33	53.01	65.95	51.63			68.32	65.95	53.01	67.11
	Linear SVM	53.28			44.82	80.14	35.79	57.49			73.47	35.79	80.14	48.13
	Non-linear SVM	60.58			Null	0.00	100.00	0.00			60.58	100.00	0.00	75.45
	ANN	61.54			54.30	15.40	91.57	23.99			62.45	91.57	15.40	74.26
LBP	LR	74.29			69.32	62.42	82.02	65.69			77.03	82.02	62.42	79.45
	kNN	70.21			66.11	50.11	83.28	57.01			71.95	83.28	50.11	77.20
	Naive Bayesian	57.71			47.78	78.28	44.32	59.34			75.82	44.32	78.28	55.94
	RF	70.27			62.16	62.84	75.10	62.50			75.64	75.10	62.84	75.37
	Linear SVM	48.17			36.83	44.02	50.87	40.10			58.27	50.87	44.02	54.32
	Non-linear SVM	60.58			Null	0.00	100.00	0.00			60.58	100.00	0.00	75.45
	ANN	71.84			67.38	55.41	82.54	60.81			73.99	82.54	55.41	78.03
GLCM	LR	67.73			59.71	55.75	75.52	57.67			72.40	75.52	55.75	73.93
	kNN	69.26			62.30	55.79	78.03	58.87			73.06	78.03	55.79	75.46
	Naive Bayesian	61.99			51.12	82.01	48.96	62.98			80.70	48.96	82.01	60.94
	RF	71.39			63.16	65.85	75.00	64.48			77.14	75.00	65.85	76.06
	Linear SVM	66.50			55.89	71.27	63.39	62.65			77.22	63.39	71.27	69.63
	Non-linear SVM	67.76			58.77	61.05	72.12	59.89			73.99	72.12	61.05	73.05
	ANN	68.69			60.64	58.65	75.22	59.63			73.65	75.22	58.65	74.43

The bold text in the table indicates the highest value of the classification result of different classifiers for the same feature.