Ghost imaging with Bayesian denoising method

Abstract

We propose a Bayesian denoising method to improve the quality of ghost imaging. The proposed method achieved the highest PSNR and SSIM in both binary and gray-scale targets with fewer measurements. Experimentally, it obtained a reconstructed image of a USAF target where the PSNR and SSIM of the image were up to 12.80 dB and 0.77, respectively, whereas those of traditional ghost images were 7.24 dB and 0.28 with 3000 measurements. Furthermore, it was robust against additive Gaussian noise. Thus, this method could make the ghost imaging technique more feasible as a practical application.