Comparison on Performance of Deep Q Network based Single-pixel Imaging Using Different Orthogonal Transformations

Zhirun WANG; Wenjing ZHAO; Aiping ZHAI; Dong WANG

doi:10.3788/gzxb20225103.0311003

Acta Photonica Sinica, Volume. 51, Issue 3, 0311003(2022)

Comparison on Performance of Deep Q Network based Single-pixel Imaging Using Different Orthogonal Transformations

Zhirun WANG¹, Wenjing ZHAO¹, Aiping ZHAI¹, and Dong WANG^1,2、*

Author Affiliations

¹College of Physics and Optoelectronics，Taiyuan University of Technology，Taiyuan 030024，China

²Key Laboratory of Advanced Transducers and Intelligent Control Systems，Ministry of Education and Shanxi Province，Taiyuan University of Technology，Taiyuan 030024，China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Spectrum of “Cameraman” image with the resolution of $64 \times 64$

Download full size

View in Article

Fig. 2. Flow of DQN-SPI

Download full size

View in Article

Fig. 3. The reconstructed "Cameraman" images using different down-sampling strategies at different sampling ratios

Download full size

View in Article

Fig. 4. The spectrum of the reconstructed "Cameraman" images using different down-sampling strategies at different sampling ratios

Download full size

View in Article

Fig. 5. Reconstruction at low sampling ratios

Download full size

View in Article

Fig. 6. $S (k)$ curve of DSPI，FSPI，HSPI and KMSPI

Download full size

View in Article

Fig. 7. Schematic of the experimental setup

Download full size

View in Article

Fig. 8. The reconstructed images of "tortoise" using different down-sampling strategies at different sampling ratios

Download full size

View in Article

Table 1. PSNR and SSIM of the reconstructed “Cameraman” images using different down-sampling strategies

View table

View in Article

Table 1. PSNR and SSIM of the reconstructed “Cameraman” images using different down-sampling strategies

		Sampling ratio
	Method	5%	10%	20%	40%
PNSR/dB	AZ-HSPI	15.04	15.80	16.43	22.06
	DQN-HSPI	19.68	21.23	23.91	27.84
	AC-FSPI	17.13	18.19	22.62	23.35
	DQN-FSPI	20.54	22.56	25.24	28.49
	AZ-DSPI	14.79	15.48	19.29	23.28
	DQN-DSPI	20.56	22.63	25.13	29.22
	AS-KMSPI	11.40	13.88	17.59	23.45
	DQN-KMSPI	14.60	18.25	22.81	26.27
SSIM /%	AZ-HSPI	45.8	57.0	70.5	82.3
	DQN-HSPI	55.5	65.2	76.6	88.2
	AC-FSPI	54.1	69.2	81.1	90.2
	DQN-FSPI	64.6	75.7	85.1	91.3
	AZ-DSPI	45.2	58.1	71.8	81.1
	DQN-DSPI	65.4	74.3	83.0	90.7
	AS-KMSPI	22.8	30.0	44.2	80.0
	DQN-KMSPI	47.6	60.0	78.6	87.0

Table 2. PSNR and SSIM of the reconstructed images of "tortoise" using different down-sampling strategies

View table

View in Article

Table 2. PSNR and SSIM of the reconstructed images of "tortoise" using different down-sampling strategies

		Sampling ratio
	Method	5%	10%	20%	40%
PNSR/dB	AZ-HSPI	18.03	19.56	21.72	24.82
	DQN-HSPI	19.87	21.95	24.20	27.41
	AC-FSPI	17.85	20.00	22.51	26.40
	DQN-FSPI	19.66	22.06	24.76	29.08
	AZ-DSPI	18.65	19.83	22.96	26.36
	DQN-DSPI	20.45	22.59	25.53	30.09
	AS-KMSPI	10.46	15.27	19.08	25.13
	DQN-KMSPI	15.21	18.07	18.63	25.76
SSIM/%	AZ-HSPI	52.0	65.7	78.7	88.6
	DQN-HSPI	70.9	80.1	86.2	91.1
	AC-FSPI	56.9	72.2	83.7	92.0
	DQN-FSPI	73.0	83.4	90.8	96.3
	AZ-DSPI	62.1	73.9	84.5	94.2
	DQN-DSPI	78.5	85.6	92.2	96.9
	AS-KMSPI	32.0	49.7	66.7	92.8
	DQN-KMSPI	69.0	80.0	83.7	92.3

Tools

Get Citation

Copy Citation Text

Zhirun WANG, Wenjing ZHAO, Aiping ZHAI, Dong WANG. Comparison on Performance of Deep Q Network based Single-pixel Imaging Using Different Orthogonal Transformations[J]. Acta Photonica Sinica, 2022, 51(3): 0311003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category:

Received: Jul. 28, 2021

Accepted: Sep. 9, 2021

Published Online: Apr. 8, 2022

The Author Email: WANG Dong (wangdong@tyut.edu.cn)

DOI:10.3788/gzxb20225103.0311003

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology