Sparse Representation Classification Based on Flexible Patches Sampling of Superpixels for Hyperspectral Images

Table 3

The classification accuracy (average on ten runs with 10% random samples) for India Pines image on six test approaches.

class	PSRC	JSRM	MJSR	MASR	BTC-WSL	SRC-FPSS
C1	39.84	84.55	83.74	94.31	99.07	92.68
C2	54.29	96.42	93.64	98.19	93.02	97.85
C3	52.21	92.41	93.44	97.50	89.91	97.90
C4	41.31	92.64	88.41	96.09	90.92	95.77
C5	85.36	94.33	94.41	96.24	94.32	98.01
C6	89.75	93.71	97.92	99.95	100.00	99.89
C7	74.67	85.33	84.00	100.00	98.15	100.00
C8	96.28	99.85	99.77	99.85	100.00	99.92
C9	24.07	31.48	59.26	66.67	100.00	70.37
C10	69.37	91.70	93.22	95.47	92.26	98.10
C11	70.88	96.73	97.13	98.98	99.17	98.87
C12	39.76	92.57	88.23	94.38	99.19	95.57
C13	91.85	80.98	93.84	98.55	100.00	98.37
C14	90.48	98.54	99.56	100.00	98.92	99.97
C15	41.11	91.64	89.15	97.60	95.48	95.39
C16	88.10	88.49	86.90	96.43	99.20	96.43
OA	69.29	94.77	94.16	98.03	96.43	98.29
AA	65.58	88.21	91.05	95.64	96.85	95.94
Kappa	64.94	94.03	94.31	97.75	95.91	98.05