DBDNet for denoising in ESPI wrapped phase patterns with high density and high speckle noise

Jianming Li; Chen Tang; Min Xu; Zirui Fan; Zhenkun Lei

doi:10.1364/AO.442293

Applied Optics
Vol. 60,
Issue 32,
pp. 10070-10079
(2021)
•https://doi.org/10.1364/AO.442293

DBDNet for denoising in ESPI wrapped phase patterns with high density and high speckle noise

Jianming Li, Chen Tang, Min Xu, Zirui Fan, and Zhenkun Lei

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Jianming Li, Chen Tang, Min Xu, Zirui Fan, and Zhenkun Lei, "DBDNet for denoising in ESPI wrapped phase patterns with high density and high speckle noise," Appl. Opt. 60, 10070-10079 (2021)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Check for updates

Related Topics
Table of Contents Category
- Imaging Systems, Image Processing, and Displays
Optics & Photonics Topics
?

The topics in this list come from the Optics and Photonics Topics applied to this article.

About this Article
History
- Original Manuscript: September 1, 2021
- Revised Manuscript: October 8, 2021
- Manuscript Accepted: October 8, 2021
- Published: November 2, 2021

Abstract

In this paper, we propose a dilated-blocks-based deep convolution neural network, named DBDNet, for denoising in electronic speckle pattern interferometry (ESPI) wrapped phase patterns with high density and high speckle noise. In our method, the proposed dilated blocks have a specific sequence of dilation rate and a multilayer cascading fusion structure, which can better improve the effect of speckle noise reduction, especially for phase patterns with high noise and high density. Furthermore, we have built an abundant training dataset with varieties of densities and noise levels to train our network; thus, the trained model has a good generalization and can denoise ESPI wrapped phase in various circumstances. The network can get denoised results directly and does not need any pre-process or post-process. We test our method on one group of computer-simulated ESPI phase patterns and one group of experimentally obtained ESPI phase patterns. The test images have a high degree of speckle noise and different densities. We compare our method with two representative methods in the spatial domain and frequency domain, named oriented-couple partial differential equation and windowed Fourier low pass filter (LPF), and a method based on deep learning, named fast and flexible denoising convolutional neural network (FFDNet). The denoising performance is evaluated quantitatively and qualitatively. The results demonstrate that our method can reduce high speckle noise and restore the dense areas of ESPI phase patterns, and get better results than the compared methods. We also apply our method to a series of phase patterns from a dynamic measurement and get successful results.

Full Article | PDF Article

More Like This

Uneven wrapped phase pattern denoising using a deep neural network

Jianming Li, Chen Tang, Min Xu, and Zhenkun Lei
Appl. Opt. 61(24) 7150-7157 (2022)

Application of attention-DnCNN for ESPI fringe patterns denoising

Linlin Wang, Run Li, Feng Tian, and Xiaoyu Fang
J. Opt. Soc. Am. A 39(11) 2110-2123 (2022)

UN-PUNet for phase unwrapping from a single uneven and noisy ESPI phase pattern

Hongxuan He, Chen Tang, Liao Zhang, Min Xu, and Zhenkun Lei
J. Opt. Soc. Am. A 40(10) 1969-1978 (2023)

Previous Article Next Article

Data Availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (7)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (5)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (8)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Methods	Input	Output	Network Architecture	Skip Connection	Dilated Convolution	Training Dataset
Hao et al. [22]	Noisy fringe patterns	Clear fringe patterns	FFDNet-based: $D o w n s a m p l i n g + 20 \times (C o n v + B N + R e l u) + U p s a m p l i n g$	No	No	$69600 - 40 \times 40$
Yan et al. [26]	Noisy sine/cosine fringe patterns	Clear sine/cosine fringe patterns	ResNet-based: $2 D - C o n v + 7 \times R e s N e t s (4 r e s i d u a l b l o c k s f o r$ $e a c h R e s N e t) + 2 D - C o n v$	Yes	No	$4, 4000 - 40 \times 40$
Yan et al. [27]	Noisy sine/cosine fringe patterns	Clear sine/cosine fringe patterns	DnCNNs-based: $35 \times (C o n v + B N + R e l u) + C o n v$	No	No	$3, 0000 - 80 \times 80$
Proposed method	Noisy phase patterns	Clear phase patterns	$D o w n s a m p l i n g + 2 \times (C o n v + B N + R e l u) + 4 \times d i l a t i o n b l o c k s + U p s a m p l i n g$	Yes	Yes	15,7000 $(435 \times 361) - 50 \times 50$

Layer	Feature Map	Padding	Dilation	Layer	Feature Map	Padding	Dilation
Downsampling	$50 \times 1 \times 4$	1	1	DB3
$C o n v + B N + R e l u$	$50 \times 4 \times 64$	1	1	$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1
DB1				$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2
$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1	$D 5 + B N + R e l u$	$50 \times 64 \times 64$	5	5
$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2	$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2
$D 5 + B N + R e l u$	$50 \times 64 \times 64$	5	5	$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1
$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2	Fusion3	$50 \times 320 \times 64$	1	1
$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1	DB4
Fusion1	$50 \times 320 \times 64$	1	1	$D 1 + B N + R e l u$	$50 \times 128 \times 64$	1	1
DB2				$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2
$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1	$D 5 + B N + R e l u$	$50 \times 64 \times 64$	5	5
$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2	$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2
$D 5 + B N + R e l u$	$50 \times 64 \times 64$	5	5	$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1
$D 2 + B N + R e l u$	$50 \times 64 \times 64$	2	2	Fusion4	$50 \times 320 \times 64$	1	1
$D 1 + B N + R e l u$	$50 \times 64 \times 64$	1	1	$C o n v + B N + R e l u$	$50 \times 128 \times 4$	1	1
Fusion2	$50 \times 320 \times 64$	1	1	Upsampling	$50 \times 4 \times 1$	1	1

Number of Interpolation Points	$n_{0}$	Number of Phase Images Generated	Total Number of Dataset 1
4	30	50	300
9	30	50
25	30	50
49	20	50
81	20	50
225	15	50

Quality Index	PSNR	SSIM
WFLPF	12.7554	0.6459
OCPDE	14.0865	0.6146
FFDNet	17.3057	0.8489
Proposed method	20.5648	0.9239

Methods	Input	Output	Network Architecture	Skip Connection	Dilated Convolution	Training Dataset
Hao et al. [22]	Noisy fringe patterns	Clear fringe patterns	FFDNet-based: $D o w n s a m p l i n g + 20 \times (C o n v + B N + R e l u) + U p s a m p l i n g$	No	No	$69600 - 40 \times 40$
Yan et al. [26]	Noisy sine/cosine fringe patterns	Clear sine/cosine fringe patterns	ResNet-based: $2 D - C o n v + 7 \times R e s N e t s (4 r e s i d u a l b l o c k s f o r$ $e a c h R e s N e t) + 2 D - C o n v$	Yes	No	$4, 4000 - 40 \times 40$
Yan et al. [27]	Noisy sine/cosine fringe patterns	Clear sine/cosine fringe patterns	DnCNNs-based: $35 \times (C o n v + B N + R e l u) + C o n v$	No	No	$3, 0000 - 80 \times 80$
Proposed method	Noisy phase patterns	Clear phase patterns	$D o w n s a m p l i n g + 2 \times (C o n v + B N + R e l u) + 4 \times d i l a t i o n b l o c k s + U p s a m p l i n g$	Yes	Yes	15,7000 $(435 \times 361) - 50 \times 50$

Abstract

Data Availability

Cited By

Figures (7)

Tables (5)

Equations (8)

Applied Optics