Division-of-Focal-Plane Polarization Image Denoising Algorithm Based on Improved Principal Component Analysis

Jiaqi Yin; Shiyong Wang; Fanming Li

doi:10.3788/AOS202141.0710002

Acta Optica Sinica, Volume. 41, Issue 7, 0710002(2021)

Division-of-Focal-Plane Polarization Image Denoising Algorithm Based on Improved Principal Component Analysis

Jiaqi Yin^1,2,3, Shiyong Wang^1,2、*, and Fanming Li^1,2、**

¹Key Laboratory of Infrared System Detection and Imaging Technology, Chinese Academy of Sciences, Shanghai 200083, China

²Shanghai Institute of Technical Physics of the Chinese Academy of Sciences, Shanghai 200083, China

³University of Chinese Academy of Sciences, Beijing 100049, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. Schematic diagram of the micro-polarizer array

Download full size

Fig. 2. Flow chart of our algorithm

Download full size

Fig. 3. Denoising model of the DoFP polarization image

Download full size

Fig. 4. DoFP polarization image obtained by simulation. (a) DoFP; (b) 0°; (c) 45°; (d) 90°; (e) 135°

Download full size

Fig. 5. Test images. (a) Fabrics; (b) leaves; (c) macbeth classic; (d) macbeth enhancement; (e) painting; (f) potery

Download full size

Fig. 6. Potery images before and after denoising. (a) Original image; (b) demosaicing image; (c) noisy image; (d) denoised image

Download full size

Fig. 7. Denoising results under different σ. (a) PSNR; (b) SSIM

Download full size

Fig. 8. Denoising results of different algorithms. (a) Original image; (b) demosaicing image; (c) noisy image; (d) PCA algorithm; (e) K-SVD algorithm; (f) BM3D algorithm; (g) our algorithm

Download full size

Fig. 9. Denoising results of different algorithms for image macbeth enhancement. (a) PSNR; (b) SSIM

Download full size

Fig. 10. Denoising result of real images by different algorithms. (a) Original image; (b) non-uniformity corrected image; (c) PCA algorithm; (d) K-SVD algorithm; (e) BM3D algorithm; (f) our algorithm

Download full size

Table 1. Stokes vector images and DoLP images before and after denoising

View table

Table 1. Stokes vector images and DoLP images before and after denoising

Image		S0	S1	S2	DoLP
Demosaicing image	PSNR /dB	46.5230	38.6166	41.6959	31.3399
	SSIM	0.9956	0.9512	0.9645	0.7770
Noisy image	PSNR /dB	35.0436	26.1517	26.2277	5.9415
	SSIM	0.8104	0.2388	0.2196	0.1175
Denoised image	PSNR /dB	41.4074	37.1018	39.5708	25.9634
	SSIM	0.9809	0.9060	0.9215	0.5934

Table 2. Average running time of different algorithms
View table
Table 2. Average running time of different algorithms
Algorithm PCA K-SVD BM3D Ours
Time /s 5.9247 34.2411 0.3866 0.8869

Table 3. Denoising results of different algorithms

View table

Table 3. Denoising results of different algorithms

Algorithm		Fabrics	Leaves	Macbethclassic	Macbethenhancement	Painting	Potery
Noisy image	PSNR /dB	28.1379	28.1232	28.1315	28.1443	28.1412	28.1259
	SSIM	0.6787	0.4759	0.6143	0.7773	0.7606	0.5223
PCA	PSNR /dB	34.5693	37.5385	38.4190	38.1194	33.8570	37.0131
	SSIM	0.9095	0.9011	0.9481	0.9734	0.9390	0.9124
K-SVD	PSNR /dB	36.1498	39.0277	37.1737	37.7568	33.6660	38.0791
	SSIM	0.9403	0.9365	0.9068	0.9608	0.9334	0.9438
BM3D	PSNR /dB	33.1474	37.7663	39.3794	37.6404	32.7475	37.3816
	SSIM	0.8937	0.9286	0.9611	0.9668	0.9316	0.9456
Ours	PSNR /dB	36.9616	41.5592	43.7434	42.5957	35.2164	40.3545
	SSIM	0.9543	0.9778	0.9900	0.9940	0.9626	0.9826

Table 4. Denoising results of different algorithms on macbeth enhancement images

View table

Table 4. Denoising results of different algorithms on macbeth enhancement images

Algorithm		S0	S1	S2	DoLP
Demosaicing	PSNR /dB	49.4055	41.9600	43.4291	33.9477
	SSIM	0.9975	0.9600	0.9816	0.8999
Noisy image	PSNR /dB	35.1736	26.2301	26.2683	11.8090
	SSIM	0.8235	0.2948	0.3727	0.1553
PCA	PSNR /dB	41.8979	35.6964	35.6833	18.5476
	SSIM	0.9723	0.8111	0.8540	0.4134
K-SVD	PSNR /dB	41.6852	34.8183	33.6769	12.9843
	SSIM	0.9744	0.7815	0.8142	0.4154
BM3D	PSNR /dB	40.9549	36.5001	35.7293	20.7019
	SSIM	0.9689	0.8720	0.8902	0.5247
Ours	PSNR /dB	43.0844	39.9698	39.6139	27.0429
	SSIM	0.9847	0.9305	0.9527	0.7623

Tools

Get Citation

Copy Citation Text

Jiaqi Yin, Shiyong Wang, Fanming Li. Division-of-Focal-Plane Polarization Image Denoising Algorithm Based on Improved Principal Component Analysis[J]. Acta Optica Sinica, 2021, 41(7): 0710002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites