Compact speckle spectrometer based on CNN-LSTM denoising.

Area of Science:

• Optical Engineering
• Materials Science
• Spectroscopy

Background:

• Speckle spectrometers offer compact spectral analysis.
• Traditional methods struggle with noise and stability.
• Femtosecond laser processing enables advanced micro-nanostructures.

Purpose of the Study:

• To develop a compact and stable speckle spectrometer.
• To improve spectral reconstruction accuracy and speed.
• To mitigate environmental noise effects on system performance.

Main Methods:

• Utilizing micro-nanostructures fabricated by femtosecond lasers on sapphire as scattering media.
• Implementing a convolutional long short-term memory network (CNN-LSTM) for spectral denoising and reconstruction.
• Evaluating spectral resolution at 0.5 nm.

Main Results:

• Achieved a spectral resolution of 0.5 nm.
• Demonstrated rapid and accurate spectral reconstruction using the CNN-LSTM algorithm.
• Significantly reduced reconstruction errors caused by environmental noise and speckle autocorrelation reduction.
• Prolonged system stability time.

Conclusions:

• The proposed compact speckle spectrometer offers high spectral resolution and stability.
• The CNN-LSTM denoising algorithm enhances reconstruction accuracy and speed.
• This technology provides a robust solution for spectral analysis in noisy environments.