Hybrid Low-Rank Tensor CP and Tucker Decomposition with Total Variation Regularization for HSI Noise Removal

Table 1

All algorithms compare the performance of WDC data set under different noise types.

Noise type	Index	Noisy	LRTV	WTV-HMCT	BM4D	WGLRTD	KBR	LLRT
Case 1	MPSNR	28.230	31.179	38.431	37.049	37.610	36.874	40.021
	MSSIM	0.869	0.940	0.987	0.981	0.982	0.980	0.989
	MFSIM	0.941	0.959	0.992	0.987	0.989	0.990	0.992
	SAM	0.215	0.117	0.057	0.069	0.062	0.049	0.042
Case 2	MPSNR	18.589	25.162	31.459	30.255	31.959	31.102	32.623
	MSSIM	0.521	0.779	0.938	0.921	0.946	0.935	0.949
	MFSIM	0.776	0.875	0.968	0.949	0.963	0.962	0.969
	SAM	0.484	0.163	0.112	0.120	0.101	0.065	0.074
Case 3	MPSNR	8.135	20.241	23.151	23.651	25.469	24.515	25.839
	MSSIM	0.109	0.449	0.740	0.697	0.791	0.742	0.813
	MFSIM	0.490	0.694	0.878	0.832	0.887	0.846	0.895
	SAM	0.949	0.213	0.216	0.180	0.199	0.089	0.109