Improving PSNR and computational efficiency in orthogonal ghost imaging

In this paper, we present an orthogonal ghost imaging (OGI) method based on two-dimensional discrete cosine transform (2D-DCT) patterns. Unlike traditional methods that are based on random or sinusoidal patterns, our method relies on structured orthogonal patterns to enhance both image quality and reconstruction speed, outperforming random and sinusoidal-based approaches in terms of reconstruction fidelity and computational efficiency. A new reconstruction formula is derived in our approach. In addition, using a phase-shift illumination pattern technique helps to effectively reduce environmental noise. Simulation and experimental results show that high-quality image reconstruction is achievable even with reduced sampling rates. For instance, using only 30% of the measurements is enough to meet the Peak Signal-to-Noise Ratio (PSNR) threshold predicted by Shannon entropy. Compared to differential and sinusoidal ghost imaging techniques, the proposed method consistently outperforms them in terms of signal-to-noise ratio (SNR) and reconstruction efficiency. These findings suggest that OGI offers a promising direction for efficient and low-cost ghost imaging systems.