Coded aperture compression temporal imaging based on a dual-mask and deep denoiser.

Area of Science:

• Optics and Photonics
• Image Processing
• Computational Imaging

Background:

• Coded aperture compressive temporal imaging (CACTI) enables high-speed photography using low-speed cameras by encoding multiple frames into a single measurement.
• Both the encoding pattern and reconstruction algorithm are critical for CACTI performance.
• Existing methods face limitations in reconstruction quality.

Purpose of the Study:

• To enhance the reconstruction quality of CACTI.
• To develop an improved encoding strategy and a novel reconstruction algorithm for CACTI.

Main Methods:

• Utilized the reflective properties of a digital micromirror device to implement a complementary dual-mask encoding pattern for richer projection information.
• Developed a novel reconstruction algorithm combining weighted Landweber regularization, a relaxation strategy, and a deep denoiser.
• Conducted experimental validation to assess the performance of the proposed encoding-decoding combination.

Main Results:

• The proposed complementary dual-mask pattern effectively captured more projection information.
• The combined weighted Landweber regularization, relaxation strategy, and deep denoiser algorithm demonstrated superior reconstruction capabilities.
• Experimental results showed significant improvements in peak Signal-to-Noise Ratio (SNR), structural similarity index measure (SSIM), and visual quality compared to existing methods.

Conclusions:

• The developed encoding-decoding strategy significantly improves CACTI performance.
• The novel approach offers a promising solution for high-speed imaging applications requiring high-quality reconstructions.
• This work advances the field of compressive imaging by enhancing both data acquisition and signal recovery.