A high performance deep learning based single-pixel image reconstruction framework integrating the U-Net structure with improved Mamba modules

Key Points

• Improved image reconstruction accuracy is achieved using the U-Net architecture, enhancing quality.
• The neural network integrates Mamba modules, optimizing performance in single-pixel images.
• Observational analysis within testing environments demonstrates the framework's capability with various images.
• This advancement supports applications in medical imaging and remote sensing, paving the way for future innovations.