Search

• Analyzing low signal-to-noise challenges in current optically-sectioned structured illumination microscopy for complex surface measurement. • Introducing a fast and robust sparse domain denoising method in optically-sectioned structured illumination microscopy for high-quality complex surface measurement. • Integrating advanced image denoising technique with the HiLo principle for high-quality optical sectioning reconstruction. With rapid advancements in precision micro/nano-systems, product surfaces are becoming increasingly complex. Despite its good adaptability, optically sectioned structured illumination microscopy faces low signal-to-noise challenges for high-quality complex surface measurement, leading to increased measurement noise, outliers, and non-measurement points. To address these challenges, we introduce a novel sparse domain robust denoising method to obtain clean optically sectioned images for robust complex surface measurement. Specifically, our approach divides the denoising process into two distinct paths: the low-frequency path focuses on exploring the intrinsic sparsity and nonlocal self-similarity of the image to minimize the noise of the low-frequency signals in an artifact-free manner, whereas the complementary high-frequency path utilizes modulation-contrast-based weighting algorithm to provide clean high-frequency signals. Finally, by fusing signals from both low-frequency and high-frequency paths, we obtain clean and nearly full-resolution optically sectioned images. Measurement experiments on various test samples demonstrate that the proposed method provides good measurement efficiency and consistently high surface reconstruction quality for different surface topographies. [ABSTRACT FROM AUTHOR]