Sub-Nyquist computational ghost imaging with orthonormalized colored noise pattern

Abstract

Computational ghost imaging generally requires a large number of pattern illumination to obtain a high-quality image. The colored noise speckle pattern was recently proposed to substitute the white noise pattern in a variety of noisy environments and gave a significant signal-to-noise ratio enhancement even with a limited number of patterns. We propose and experimentally demonstrate here an orthonormalization approach based on the colored noise patterns to achieve sub-Nyquist computational ghost imaging. We tested the reconstructed image in quality indicators such as the contrast-to-noise ratio, the mean square error, the peak signal to noise ratio, and the correlation coefficient. The results suggest that our method can provide high-quality images while using a sampling ratio an order lower than the conventional methods.Computational ghost imaging (CGI) [1-3], an ameliorated scheme on traditional ghost imaging (GI) [4][5][6][7], owns the ability to reconstruct the object via single bucket detector. CGI also grants advantages in an expanding range of non-conventional applications such as wide spectrum imaging [8,9] and depth mapping [10,11]. It also finds application to various fields, such as temporal imaging [3], X-ray imaging [12], remote sensing [13], etc.. However, its sampling number is usually comparable to the total number of pixels in the speckle pattern to keep good imaging quality. Thus, it is time-consuming and resource-intensive Besides, it is only suitable for static object reconstruction.Various methods have been proposed to overcome this problem [14][15][16][17][18][19]. One typical and effective way is the orthonormalization method [18]. This method introduces a data post-processing algorithm to improve the reconstructing process in a GI system with pseudo-thermal light. The required sampling number is reduced by applying the Gram-Schmidt process on the noise patterns and intensity sequence collected by the bucket detector However, such a method is sensitive to noise, and the image quality is not even comparable with standard CGI when the pattern number is large enough. Traditionally, Gaussian white noise speckle pattern is used for GI. The spatial distribution of the light field amplitude is Gaussian, and the phase associated with the field amplitude is random. We recently developed a method to generate the so-called colored noise speckle pattern for CGI by customizing the speckle patterns' power spectrum density [20]. Unlike white noise, colored noise generally has non-zero cross-correlation between neighborhood pixels. Sub-Rayleigh imaging was demonstrated with the blue noise pattern, which has negative cross-correlation between two adjacent pixels. The pink noise pattern allowed us to image in a variety of noisy environments.This letter introduces a novel method for combining the colored noise and Orthonormalization methods to substantially reduce the number of sampling and overcome the drawback of pink noise patterns. We also compare the orthonormalization colored noise GI (OCGI) with orthon...

Deep-learned speckle pattern and its application to ghost imaging

“…The requirement for speckle patterns increases immeasurably when the object's pixel size is large. Many ameliorated techniques have been proposed to minimize the SR, such as orthonormalization method [30,34], Fourier and sequency Walsh-Hadamard speckles [31,32,43], and compressive sensing [44,45].…”

“…Various synthesized speckle patterns [25][26][27][28][29] are generated by customizing and regulating amplitudes and phases of the electro-magnetic fields or directly designing and adjusting the power spectrum of the speckle patterns. Recently, efforts have been made to generate orthonormalized [30], Walsh-Hadamard [31][32][33], and colored noise [34] speckle patterns for sub-Nyquist sampling imaging. To date, the synthesized speckle patterns are typically generated from customizing the power spectrum, vortex, amplitude of either the intensity or field distribution to justify their spatial correlations.…”

0.8% Nyquist computational ghost imaging via non-experimental deep learning

“…It nevertheless largely depends on the sparsity of objects and is sensitive to noise [18]. Orthonormalized noise patterns can be used to suppress the noise and improve the image's quality under a limited sampling number [19,20]. In particular, the orthonormalized colored noise patterns can break the Nyquist limit down to ∼ 5% [20].…”

“…Orthonormalized noise patterns can be used to suppress the noise and improve the image's quality under a limited sampling number [19,20]. In particular, the orthonormalized colored noise patterns can break the Nyquist limit down to ∼ 5% [20]. Fourier and sequence-ordered Walsh-Hadamard patterns, which are orthogonal to each other in time or spatial domain, were also applied to the sub-Nyquist imaging [21][22][23].…”