Compressive ghost imaging in scattering media guided by region of interest

Abstract

Compressive ghost imaging (CSGI) combines structured illumination and a bucket detector for obtaining the light intensity signal from an unknown object. Light fluctuations are generated by few measurements and then an image is reconstructed using optimization algorithms such as compressive sensing (CS) by finding its sparse representation. The measured light fluctuations are not sensitive to scattering degradation. Consequently, an absorbing object completely embedded in a scattering media can be imaged. To speed up the sequential loading of illumination patterns on a digital micromirror device and achieve data compression, the sampling number should be reduced because more than 90% of the time in CS reconstruction is usually spent in getting a frame of image. In this study, we propose a novel strategy to realize a speedy and reliable reconstruction procedure for obtaining a high image quality by using prior knowledge during the acquisition of light intensity signal in CSGI. The prior knowledge is established by extracting the features of a local target area, which is designed by descattering of images with extra noise using fast Fourier single-pixel imaging. The proposed method facilitates a reliable image quality even under the reduction in the compression ratio, thereby overcoming the limitation of the dependence of sampling ratio on the image quality in CSGI.

Tilt illumination for structured illumination imaging

“…Several pictures are generated by moving a grating for reconstructing a high resolution image [25,26,27,28]. The grating function is expressed as follows…”

Tilt illumination for structured illumination imaging

“…Several pictures are generated by moving a grating for reconstructing a high resolution image [25,26,27,28]. The grating function is expressed as follows m(x, y) = cos[2πk 0 (x cos θ + y sin θ) + α],…”

Learning Approach for Flexible Spherical/Aspherical Reflective Surface Control

“…In addition, learning-based methods have recently become a hot topic in computer science. (14) However, it usually requires many hours or even days to experimentally collect tens of thousands of in situ data points for neural network training. This large cost makes these methods unsuitable for in situ imaging, particularly in the process of wavefront measurement, which also requires stable environmental conditions to sample many variable deformable surfaces.…”

“…This method can reduce oscillations in the control parameters; however, the entire iteration process converges stably and the iteration time is relatively long. (20,21) To improve the convergence speed, an iterative control method was proposed to optimize the solution process of the IF using a multivariate statistical method (12,13) In addition, an alternative iterative measurement approach was investigated by calculating the mapping relationship (i.e., the mapping between the polynomial terms of the surface shape and actuator driving voltage), (12)(13)(14)(15)(16)(17) with the results showing that on average at least 10 iterations are needed to converge to the desired surface shape. One way to increase the control efficiency is to address the deformation of DMs in the mechanical model (22) and obtain the analytical solutions of the control parameters based on a surface shape correction.…”