Non-diffractive computational ghost imaging

Abstract

Computational ghost imaging (CGI) enables an image to be recorded using a single-pixel detector. The image can be reconstructed from correlations between the scene and a series of known projected intensity patterns. In this work we investigate the performance of CGI using pseudo non-diffracting (ND) speckle patterns. We demonstrate an extended depth-of-field that is ∼ 2-3 times greater than that achievable with conventional speckle, when only computing each intensity pattern to a single depth. In addition, the average speckle grain size of ND speckle is reduced by a factor of ∼ 1.5 relative to conventional speckle, which enhances the lateral Rayleigh-limit resolving power of our reconstructed images. However, the point-spread function (PSF) of our imaging system takes the form of a Bessel beam, which manifests itself as long-range correlations between speckle grains in the projected patterns. We discuss the trade-off between enhancement of the depth-of-field and the lateral resolution when using ND speckle, at the expense of a reduction in image contrast. Our work demonstrates that the tailoring of lateral and axial correlations in projected intensity patterns permits PSF engineering in CGI.