Table 5.

Classification results of five image features using different classifiers in the 80 × 80 pixels sub-database of GasHisSDB [In (%)].

Freatures	Methods	Acc			Abnormal						Normal
					Pre	Rec	Spe	F1			Pre	Rec	Spe	F1
	LR	82.22			78.17	77.59	85.35	77.88			84.93	85.35	77.59	85.14
	kNN	85.24			80.60	83.52	86.41	82.03			88.58	86.41	83.52	87.48
	RF	83.27			77.14	83.15	83.34	80.03			87.98	83.34	83.15	85.60
Color histogram	Linear SVM	60.81			50.86	83.58	45.41	63.24			80.36	45.41	83.58	58.03
	Non-linear SVM	59.67			Null	0.00	100.00	0.00			59.67	100.00	0.00	74.74
	ANN	79.28			76.60	70.03	85.54	73.17			80.85	85.54	70.03	83.13
	LR	70.16			66.79	51.77	82.60	58.33			71.70	82.60	51.77	76.76
	kNN	74.65			67.67	71.11	77.04	69.35			79.78	77.04	71.11	78.38
	RF	75.10			67.77	72.94	76.55	70.26			80.71	76.55	72.94	78.58
	Linear SVM	54.58			46.58	85.81	33.47	60.38			77.72	33.47	85.81	46.79
Luminance histogram	Non-linear SVM	59.67			Null	0.00	100.00	0.00			59.67	100.00	0.00	74.74
	ANN	70.17			63.19	62.38	75.43	62.78			74.79	75.43	62.38	75.11
HOG	LR	59.87			53.42	3.96	97.67	7.37			60.07	97.67	3.96	74.39
	kNN	59.63			49.95	42.22	71.40	45.76			64.64	71.40	42.22	67.85
	Naive Bayesian	55.91			46.97	72.35	44.79	56.96			70.56	44.79	72.35	54.79
	RF	59.08			49.31	51.88	63.95	50.56			66.28	63.95	51.88	65.10
	Linear SVM	53.47			44.46	61.60	47.98	51.64			64.89	47.98	61.60	55.17
	Non-linear SVM	59.70			90.91	0.08	99.99	0.17			59.68	99.99	0.08	74.75
	ANN	59.67			50.08	2.49	98.32	4.75			59.87	98.32	2.49	74.42
LBP	LR	70.92			65.32	59.49	78.65	62.27			74.17	78.65	59.49	76.34
	kNN	68.48			63.20	52.32	79.41	57.24			71.13	79.41	52.32	75.04
	Naive Bayesian	59.09			49.55	77.69	46.52	60.51			75.52	46.52	77.69	57.57
	RF	68.16			60.13	62.49	71.98	61.29			73.95	71.98	62.49	72.95
	Linear SVM	43.10			27.68	25.48	55.01	26.53			52.20	55.01	25.48	53.56
	Non-linear SVM	59.67			Null	0.00	100.00	0.00			59.67	100.00	0.00	74.74
	ANN	68.57			62.75	54.32	78.21	58.23			71.69	78.21	54.32	74.81
GLCM	LR	65.56			57.32	57.24	71.19	57.28			74.65	71.21	64.23	72.89
	kNN	68.84			62.32	57.53	76.49	59.83			72.71	76.49	57.53	74.55
	naive Bayesian	62.12			51.96	80.87	49.45	63.27			79.27	49.45	80.87	60.91
	RF	68.39			60.13	64.23	71.21	62.11			74.65	71.21	64.23	72.89
	Linear SVM	66.82			57.14	71.04	63.97	63.33			76.57	63.97	71.04	69.71
	Non-linear SVM	68.31			61.03	59.26	74.42	60.13			72.99	74.42	59.26	73.70
	ANN	65.52			56.70	61.40	68.30	58.96			72.36	68.30	61.40	70.27

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