Transport of intensity phase imaging under a low signal-to-noise ratio condition

Abstract

Two phase imaging methods with transport of intensity equation (TIE) under a low signal to noise ratio (SNR) are introduced. One is a TIE phase imaging with deep learning. It is useful for parallel TIE using a diffractive optics to produce defocus images simultaneously. The defocus images are obtained by an optical convolution integral by the calculated blurred point spread functions. However the point spread functions are different from the ideals due to the limited extent and/or limited resolution of the diffractive optics. This means that an SNR of the through-focus images is low. Therefore, deep learning compensates the error. Another is transport-of- intensity computational ghost imaging (TI-CGI). It is a combination of TIE and a computational ghost imaging (CGI). It is useful for noninvasive imaging for the biomedical field because most cells are photo-sensitive and often suffer from phototoxicity. However, CGI can obtain only amplitude information. In the biomedical field, a phase information is important to know the physical parameters. To achieve, under weak illumination, it is difficult to obtain through-focus images with high SNRs. Therefore, combination of TIE and CGI is useful.