| Single-pixel imaging is a novel computational imaging and it is characterized by capturing images by using a spatially unresolvable photodetector rather than an image sensor.Single-pixel imaging techniques are proposed to use a spatial light modulator(such as,DMD,LCo S)to generate structured patterns so as to modulate the spatial information of a scene into temporally-varying light signals and the resulting one-dimensional light signals are collected by a single-pixel detector(such as,photodiode,solar cell,and photomultiplier tube).The desired object image is recovered from the collected one-dimensional signals through a certain image reconstruction algorithm.Fourier single-pixel imaging is one of the single-pixel imaging techniques.It is characterized by using Fourier basis patterns for spatial information modulation and using an inverse Fourier transform algorithm for image reconstruction.In comparison with traditional imaging techniques,single-pixel imaging shows advantages in some special situations,such as imaging at nonvisible wavebands,imaging under weak light conditions,imaging through scattering media,etc.However,the key to single-pixel imaging is to trade temporal resolution for spatial resolution.As a result,single-pixel imaging takes a great number of single-pixel measurements to retrieve a high-quality object image and the resulting imaging time is long.In order to improve the imaging efficiency of Fourier single-pixel imaging,two different methods are proposed.One method is based on the optimization of Fourier spectrum sampling and the other is based on the human vision property.First,efficient Fourier single-pixel imaging based on Gaussian random sampling is proposed.The key to this method is to perform a density-varying sampling in the Fourier space and,more importantly,the density with respect to the importance of Fourier coefficients is subject to a one-dimensional Gaussian function.Combining the Gaussian sampling strategy and compressive sensing L1-Magic reconstruction algorithm,the proposed method can maximize the spatial information of target objects.The simulations and experiments verify that the proposed method can reconstruct a sharp and clear image of 256×256 pixels with a sampling ratio of 10%.The proposed method enables fast single-pixel imaging and provides a new approach for efficient spatial information acquisition.Second,an efficient full-color Fourier single-pixel imaging method is proposed.The method is based on the fact that human eyes have a poorer spatial resolution to blue color than red and green colors.Thus,the method is proposed to sample the Fourier coefficients of blue component of color images with an ultra-low sampling ratio,and sample the Fourier coefficients of red and green components with a relatively high sampling ratio.Both numerical simulations and experiments demonstrate this method can reduce 76.7% single-pixel measurements in full-color Fourier single-pixel imaging.This method not only generates a new insight of how to optimize the imaging efficiency by utilizing human vision properties,but also can be adopted by other full-color computational imaging techniques. |
