.png){kind=icon}
.png){kind=icon}
Return
UI-MSDE-diffusion: Fringe pattern noise elimination based on diffusion model
.png){kind=icon}
PRE
AI
DOI：10.1016/j.optlastec.2025.112435.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Optical interferometric techniques have been proven to be attractive in both research and engineering fields for non-contact, high measurement accuracy and full-field detection. Fringe patterns are the recorded results by these techniques and usually require denoising to improve the accuracy of information retrieval. Most researches on fringe pattern denoising do not take into account both uneven illumination and speckle noise. However, the uneven illumination of fringe pattern can lead to more prominent noise in certain areas, affecting the performance of denoising methods. We propose a fringe pattern denoising method called UI-MSDEdiffusion, which consists of two steps. In the first step, the fringe pattern under different lighting conditions is transferred to uniform illumination through a unified illumination network, denoted as UINet. In the second step, speckle noise in the fringe pattern is filtered out by combining the mean-reverting stochastic differential equation and the diffusion model, denoted as MSDE-diffusion. Experiments on simulated and experimental fringe patterns are performed to demonstrate the effectiveness of the proposed UI-MSDE-diffusion method.
Keywords:
Fringe pattern
Speckle noise
Uneven illumination
Diffusion model
Journal
O
IF:
5
Papers:
1.9K
Citations:
3.5W