Computational ghost imaging with spatiotemporal encoding pseudo-random binary patterns.

Area of Science:

• Optics
• Information Theory
• Computational Imaging

Background:

• Computational ghost imaging (CGI) reconstructs images without lenses or sensors.
• Existing CGI systems primarily utilize spatial patterns, overlooking temporal or space-time encoding possibilities.
• Random illumination patterns in CGI, despite noise, are advantageous for optical security applications.

Purpose of the Study:

• To explore the potential of random lighting patterns in the space-time dimension for large-scale information embedding within CGI.
• To develop novel pseudo-random pattern generation schemes for enhanced data embedding in CGI.

Main Methods:

• Inspired by binary CGI and second-order correlation, two novel pseudo-random pattern generation schemes were designed.
• 10,000 ghost images (64x64 pixels) of Hadamard-matrix-based data container patterns were embedded within the CGI framework.
• Quantitative decoding of ghost images was performed to recover standard grayscale images.

Main Results:

• Successfully embedded a total data volume of 1,280,000 bits, decoded into two 8-bit grayscale images.
• The proposed scheme demonstrated good noise resistance and a low symbol error rate.
• The information capacity and number of lighting patterns can be adjusted based on accuracy-efficiency trade-offs.

Conclusions:

• The developed space-time encoding scheme significantly enhances information embedding capacity in CGI.
• This research opens new avenues for CGI applications in optical encryption, authentication, and watermarking.
• The flexibility in pattern design allows for tailored solutions balancing accuracy and efficiency.