A wavelet denoising approach based on unsupervised learning model

EURASIP Journal on Advances in Signal Processing

Table 2 The SSIM results of the denoised images by different denoising schemes

Image	Method	σ=10	σ=25	σ=50	σ=70	σ=100
Barbara	Wavelet thresholding	0.8155	0.6022	0.4774	0.4185	0.2992
	TV	0.8283	0.5996	0.5035	0.4997	0.4615
	K-SVD	0.8996	0.7991	0.6410	0.5355	0.4453
	BM3D	0.9112	0.8342	0.7193	0.6388	0.5390
	Proposed approach	0.9608	0.9030	0.7938	0.7104	0.5977
House	Wavelet thresholding	0.7100	0.5564	0.5012	0.4368	0.3518
	TV	0.8125	0.7503	0.6783	0.6423	0.5865
	K-SVD	0.8446	0.7555	0.6667	0.5856	0.4970
	BM3D	0.8687	0.7755	0.7120	0.6628	0.5914
	Proposed approach	0.9536	0.9044	0.8362	0.7738	0.6721
Flinstones	Wavelet thresholding	0.8088	0.6614	0.4880	0.4151	0.3325
	TV	0.8531	0.7568	0.5534	0.4814	0.4591
	K-SVD	0.8697	0.7903	0.6935	0.6094	0.4830
	BM3D	0.8686	0.8015	0.7272	0.6713	0.5920
	Proposed approach	0.9391	0.8786	0.7978	0.7311	0.6294
Bridge	Wavelet thresholding	0.8492	0.6197	0.3855	0.3141	0.2718
	TV	0.8588	0.6824	0.5184	0.4204	0.3207
	K-SVD	0.8813	0.6770	0.4642	0.3864	0.3271
	BM3D	0.8900	0.7013	0.5161	0.4402	0.3710
	Proposed approach	0.9252	0.7870	0.6394	0.5618	0.4783
Fingerprint	Wavelet thresholding	0.9494	0.8462	0.6506	0.5507	0.4125
	TV	0.9460	0.8620	0.7780	0.7084	0.3747
	K-SVD	0.9647	0.8919	0.7411	0.5892	0.4160
	BM3D	0.9636	0.8950	0.8043	0.7432	0.6600
	Proposed approach	0.9701	0.9112	0.8276	0.7678	0.6690

For each test setting, five results are provided: wavelet thresholding, TV, K-SVD, BM3D, and our proposed model. The highest SSIM values (best denoising results) are given in bold