Spatial heterodyne Raman spectroscopy is a powerful tool for high-resolution spectral detection but is often hindered by weak signals, noise, and complex baselines caused by instrumental and environmental factors. For conventional spectral processing methods, it is difficult to solve noise and baseline distortion issues simultaneously without human intervention.

In a study published in Optics and Laser Technology, a team led by Prof. WANG Quan from the Xi'an Institute of Optics and Precision Mechanics of the Chinese Academy of Sciences proposed a multi-level signal enhancement network model, InDNet, which simultaneously achieves denoising and baseline correction.

Researchers developed a comprehensive interferogram simulation model that generates realistic training data. They integrated the proposed InDNet with multi-scale local feature extraction and global context modeling via Transformer blocks. Guided by a multidimensional gradient-consistent regularization loss, InDNet enhanced the structural consistency in interferograms and improved the spectral recovery.

The InDNet achieved Structural Similarity Index values of 0.9757 and 0.9827 on simulated and real data, respectively, significantly outperforming techniques such as Pyramid Detection Network and Lightweight Residual Dense U-net. It also excelled in recovering weak signals with signal-to-noise ratio ≤ 3, showing strong potential in fields such as biomedical sensing, material analysis, and chemical detection.

This study provides a robust solution for interference data enhancement, and a framework which can be applied to various high-resolution spectroscopic systems. It opens the door to fully automated, high-precision spectral processing without manual parameter tuning.