Adaptive DSPI phase denoising using mutual information and 2D variational mode decomposition

被引：9

作者：

Xiao, Qiyang ^{[1
]}

Li, Jian ^{[1
]}

Wu, Sijin ^{[2
]}

Li, Weixian ^{[2
]}

Yang, Lianxiang ^{[2
,3
]}

Dong, Mingli ^{[2
]}

Zeng, Zhoumo ^{[1
]}

机构：

[1] Tianjin Univ, State Key Lab Precis Measurement Technol & Instru, Tianjin 300072, Peoples R China

[2] Beijing Informat Sci & Technol Univ, Sch Instrumentat Sci & Optoelect Engn, Beijing 100192, Peoples R China

[3] Oakland Univ, Dept Mech Engn, Rochester, MI 48309 USA

来源：

MEASUREMENT SCIENCE AND TECHNOLOGY | 2018年 / 29卷 / 04期

关键词：

digital speckle pattern interferometry; mutual information; denoising; 2D-VMD; SPECKLE PATTERN INTERFEROMETRY; IMAGE DECOMPOSITION; FRINGE PATTERNS; RETRIEVAL; TRANSFORM;

D O I：

10.1088/1361-6501/aaa380

中图分类号：

T [工业技术];

学科分类号：

08 ;

摘要：

In digital speckle pattern interferometry (DSPI), noise interference leads to a low peak signal-to-noise ratio (PSNR) and measurement errors in the phase map. This paper proposes an adaptive DSPI phase denoising method based on two-dimensional variational mode decomposition (2D-VMD) and mutual information. Firstly, the DSPI phase map is subjected to 2D-VMD in order to obtain a series of band-limited intrinsic mode functions (BLIMFs). Then, on the basis of characteristics of the BLIMFs and in combination with mutual information, a self-adaptive denoising method is proposed to obtain noise-free components containing the primary phase information. The noise-free components are reconstructed to obtain the denoising DSPI phase map. Simulation and experimental results show that the proposed method can effectively reduce noise interference, giving a PSNR that is higher than that of two-dimensional empirical mode decomposition methods.

引用

页数：11