|M.Sc Student||Lahav Michal|
|Subject||Blind Deconvolution in 3D Biological Microscopy|
|Department||Department of Electrical Engineering||Supervisor||Professor Emeritus Arie Feuer|
|Full Thesis text|
Deconvolution of 3D fluorescent microscopy images using computational restoration techniques has attracted great interest during the last decades. Fluorescence microscope imaging properties distort the original 3D image and reduce the maximal resolution obtainable by the imaging system. Tow main factors are responsible to the degradation in image quality and resolution: One is blurring, caused by out of focus light that reaches the detectors and by diffraction phenomena that occurs in the microscope. This blur is mathematically modeled by the point spread function (PSF) of the microscope. The second factor is noise, which is caused mainly by the quantum nature of the photons detected to produce the image. The noise is mathematically described using a statistical model of an inhomogeneous Poisson point process. Deconvolution techniques attempt to mitigate these distortions, by assuming prior knowledge on the point spread function of the microscope. Blind deconvolution methods attempt to restore the original image without assuming complete knowledge on the point spread function. The purpose of this paper is to review the restoration methods, both blind and non blind, known in the field of fluorescent microscopy. The review starts by describing the principles of imaging techniques, the main factors that affect image quality and the mathematical models describing the degradation. Subsequently, the restoration methods are described. It then goes on to compare the results obtained from selected methods by means of simulations constructed for this purpose.