Sparse Fourier single-pixel imaging

Area of Science:

• Optics
• Image Processing
• Computational Imaging

Background:

• Fourier single-pixel imaging (FSPI) is a key technique.
• Current FSPI methods reduce samples by discarding high-frequency data, sacrificing image detail and resolution.
• Frequency truncation in existing methods causes significant ringing artifacts.

Purpose of the Study:

• To develop a sparse FSPI method that reduces sample acquisition while preserving and enhancing image quality.
• To address the limitations of traditional low-frequency sampling in FSPI.
• To achieve high-resolution imaging beyond diffraction limits.

Main Methods:

• A sparse Fourier single-pixel imaging approach utilizing variable density random sampling.
• Exploiting the natural decay of information power from low to high frequencies in Fourier spectra.
• Applying compressive sensing algorithms to reconstruct high-quality images from sparse Fourier data.

Main Results:

• The proposed method significantly improves object restoration quality compared to methods using only low-frequency components.
• Demonstrated effectiveness in reducing sample requirements while maintaining high image fidelity.
• Achieved super-resolution imaging capabilities, leveraging the system's diffraction-limited resolution.

Conclusions:

• The novel sparse FSPI method effectively balances sample reduction with enhanced image quality and resolution.
• Experimental validation confirms the method's correctness and practical effectiveness.
• This technique offers a promising advancement for single-pixel imaging applications.