Spatial Domain Terahertz Image Reconstruction Based on Dual Sparsity Constraints

Area of Science:

• Physics
• Imaging Science
• Signal Processing

Background:

• Terahertz time domain spectroscopy (THz-TDS) imaging systems face challenges with long acquisition times and large data volumes.
• Reducing sampling rates in THz imaging degrades reconstruction quality.
• Existing reconstruction models often struggle to balance acquisition speed and image fidelity.

Purpose of the Study:

• To propose a novel terahertz imaging model, the dual sparsity constraints terahertz image reconstruction model (DSC-THz).
• To enhance the quality of terahertz image reconstruction, particularly at reduced sampling rates.
• To address the trade-off between imaging speed and reconstruction accuracy in THz imaging.

Main Methods:

• Developed the DSC-THz model, integrating sparsity constraints from both wavelet and gradient domains.
• Introduced a non-linear exponentiation transform for wavelet coefficients to amplify significant features and suppress noise.
• Employed the split Bregman iteration scheme for efficient optimization, decomposing the problem into solvable subproblems.

Main Results:

• The DSC-THz model demonstrated superior terahertz image reconstruction quality compared to conventional single sparsity constraint models.
• Experimental results confirmed the effectiveness of the dual sparsity approach at low sampling rates.
• The proposed method successfully preserved image edges while enhancing overall sparsity.

Conclusions:

• The DSC-THz model offers a significant advancement in terahertz imaging reconstruction.
• This approach effectively mitigates the degradation in image quality associated with lower sampling rates.
• The fusion of dual sparsity constraints provides a robust solution for faster and higher-quality terahertz imaging.