White-light interferometry denoising method based on variational mode decomposition for high-precision surface topography analysis

• A novel denoising method for white-light interferometry signals is proposed, combining variational mode decomposition with an improved wavelet thresholding mechanism. • The method's effectiveness is validated through experiments on standard step, silicon wafers and gratings using a developed white-light interferometry system. • Comparative analysis demonstrates superior denoising performance against existing methods. White-light interferometry is widely used in precision manufacturing and semiconductor fabrication due to its high precision and non-contact characteristics. However, noise in the interferometric signals significantly affects measurement accuracy. This paper proposes a denoising mechanism based on variational mode decomposition (VMD) combined with improved wavelet thresholding to suppress signal noise. The simulations demonstrate that this method can effectively improve the signal-to-noise ratio, thereby enhancing signal quality and surface morphology measurement accuracy. Compared to unprocessed measurements, the measurement error is reduced by 60%. The measurements on silicon wafers (repeatability error <5 nm) and gratings (height error 1.14%) demonstrate the algorithm's consistent performance under these tested conditions. Additionally, comparisons with existing denoising algorithms further validate the advantages of the proposed method.