Multi-module collaborative optimization-driven fast speckle correlation imaging in variable environments.

Area of Science:

• Optical imaging
• Image reconstruction
• Scattering media imaging

Background:

• Speckle correlation imaging offers non-invasive imaging through scattering media via the optical memory effect.
• Practical implementation is hindered by slow reconstruction speeds.
• Developing faster algorithms is crucial for broader applications.

Purpose of the Study:

• To present a fast speckle correlation imaging (FSCI) algorithm to overcome reconstruction speed limitations.
• To enhance imaging quality and computational efficiency in scattering media.
• To demonstrate the algorithm's applicability in challenging conditions.

Main Methods:

• Developed a Fast Speckle Correlation Imaging (FSCI) algorithm with three innovations: intelligent sub-speckle quality control and initialization adaptation (ISQIA), weighted coherent averaging (WCA), and dynamic iteration termination (DIT).
• ISQIA includes sub-speckle screening (SS) and adaptive initialization (AI) for optimized convergence.
• WCA uses cross-correlation peaks for weighting, and DIT halts computation when changes are negligible.

Main Results:

• The FSCI algorithm is 40.9% faster than traditional methods, achieving reconstruction in 3.13 s.
• Significant improvements in imaging quality were observed: PSNR reached 27.04 dB and SSIM improved to 0.87.
• The algorithm demonstrated stable performance under varying noise levels (SNR 1.79 to -5.78 dB) and successfully recovered complex structures.

Conclusions:

• The FSCI algorithm offers a substantial improvement in speed and quality for imaging through scattering media.
• Its robustness in noisy conditions and ability to reconstruct complex details highlight its practical utility.
• FSCI holds significant potential for biomedical imaging and industrial non-destructive testing.