Hyperspectral Image Classification Using Dual-Branch Residual Networks

Fig. 6. Classification results of IP dataset. (a) Pseudo-color image; (b) true class; (c) DSSRN; (d) ACSS-GCN; (e) HybridSN; (f) CDC-MDAA; (g) SpectralNET; (h) Tri_CNN

Download full size

Fig. 7. Classification results of PU dataset. (a) Pseudo-color image; (b) true class; (c) DSSRN; (d) ACSS-GCN; (e) HybridSN; (f) CDC-MDAA; (g) SpectralNET; (h) Tri_CNN

Download full size

Fig. 8. Classification results of KSC dataset. (a) Pseudo-color image; (b) true class; (c) DSSRN; (d) ACSS-GCN; (e) HybridSN; (f) CDC-MDAA; (g) SpectralNET; (h) Tri_CNN

Download full size

Table 1. Information on IP, PU,and KSC datasets
View table
Table 1. Information on IP, PU,and KSC datasets
Dataset Size /（pixel×pixel） Spatial resolution /m Band Wavelength coverage /nm Kind
IP 145×145 20.0 200 400‒2400 16
PU 610×340 1.3 103 430‒860 9
KSC 512×614 18.0 176 400‒2500 13

Table 2. Classification results for IP dataset

View table

Table 2. Classification results for IP dataset

Class	DSSRN	ACSS-GCN	HybridSN	CDC-MDAA	SpectralNET	Tri_CNN
1	100.00	100.00	94.44	97.53	76.36	100.00
2	95.57	92.81	95.38	100.00	97.59	41.83
3	99.78	93.02	87.09	99.43	98.54	42.50
4	97.19	97.97	82.35	96.64	97.16	42.59
5	96.46	95.03	93.72	95.03	95.04	99.28
6	97.53	98.87	93.86	99.23	99.84	64.30
7	100.00	100.00	86.37	99.45	83.82	0
8	99.23	100.00	99.76	82.63	100.00	82.69
9	100.00	76.19	54.84	99.74	51.43	0
10	96.83	98.60	96.38	97.31	97.93	67.28
11	98.25	93.01	95.69	97.38	97.76	54.77
12	98.63	96.19	82.95	98.05	96.89	94.66
13	97.84	96.61	99.40	98.70	97.13	100.00
14	99.36	97.39	97.82	98.42	96.47	82.33
15	97.12	96.72	96.41	87.18	97.35	82.93
16	95.54	84.54	96.30	92.63	98.48	100.00
OA	97.18	95.24	93.99	95.74	96.93	60.88
AA	95.31	93.12	89.81	94.82	94.80	37.55
Kappa	96.43	94.56	94.44	95.87	96.50	53.91

Table 3. Classification results for PU dataset

View table

Table 3. Classification results for PU dataset

Class	DSSRN	ACSS-GCN	HybridSN	CDC-MDAA	SpectralNET	Tri_CNN
1	100.00	99.33	98.91	100.00	99.17	98.89
2	99.64	99.96	99.86	99.48	100.00	95.64
3	100.00	99.30	100.00	100.00	100.00	82.93
4	99.24	97.90	98.70	100.00	99.62	98.33
5	100.00	100.00	99.83	100.00	100.00	98.74
6	100.00	100.00	98.97	100.00	100.00	99.88
7	99.83	99.17	98.75	99.39	99.09	95.76
8	99.77	98.03	96.64	99.86	99.51	81.72
9	99.35	99.26	99.25	99.65	99.14	86.53
OA	99.75	99.48	99.30	99.64	99.69	94.55
AA	99.44	98.77	98.21	99.28	99.34	93.45
Kappa	99.71	99.31	99.07	99.53	99.59	92.74

Table 4. Classification results for the KSC dataset

View table

Table 4. Classification results for the KSC dataset

Class	DSSRN	ACSS-GCN	HybridSN	CDC-MDAA	SpectralNET	Tri_CNN
1	89.77	81.65	94.06	94.27	87.65	62.30
2	98.42	67.35	90.70	90.93	89.74	41.67
3	95.78	69.62	83.98	90.27	84.36	31.82
4	96.83	59.73	81.21	65.06	36.79	42.86
5	100.00	67.92	86.06	96.83	100.00	19.69
6	94.64	79.83	84.86	90.54	71.87	14.52
7	100.00	90.63	84.91	100.00	100.00	76.56
8	99.81	70.85	85.68	96.04	93.67	43.20
9	98.66	95.92	84.42	96.74	90.63	56.27
10	100.00	53.86	91.67	98.76	92.83	25.61
11	100.00	100.00	98.92	95.83	90.67	71.10
12	100.00	63.82	79.06	80.94	92.88	44.91
13	100.00	79.95	98.35	99.89	89.71	94.03
OA	97.16	76.31	89.64	93.75	83.95	53.83
AA	95.39	76.31	89.80	92.84	77.85	44.83
Kappa	96.83	68.75	87.78	93.39	82.09	48.57

Tools

Get Citation

Copy Citation Text

Tianjiao Du, Yongsheng Zhang, Lidong Bao. Hyperspectral Image Classification Using Dual-Branch Residual Networks[J]. Laser & Optoelectronics Progress, 2024, 61(22): 2237004

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites