Phase Unwrapping Problems in Digital Holographic Microscopy of Biological Cells - Biomedical Optical Phase Microscopy and Nanoscopy - page 132

0 200 400 600 800

Pixels

Original profile with curvature
First iteration
Curvature corrected profile
Curvature corrected profile
(with height adjustment)

Figure 7.6
Background subtraction by modified polyfit method.

of this modified algorithm applied to the simulated cells in Figure 7.5A is presented in
Figure 7.5D .

7.3 Varying Reconstruction Distance Method

In a case when the second wavelength is not available, the software unwrapping can be
accomplished by varying the reconstruction distance algorithm [8] . Since the resolution of
the digital hologram in XY -plane is diffraction limited to the wavelength of light, the
resolution in the z -direction is on the order of several nanometers. Thus, each pixel of a
phase map represents the average height of the object within this pixel. As a result, if one
changes the reconstruction distance by a small amount in the angular spectrum algorithm
and compares the two reconstructed fields, they appear almost identical, except the 2 π

discontinuities are shifted. Taking only the continuous parts of each image, the unwrapped
phase profile can be recreated. This “digital scanning” is possible because of many
advantages of the angular spectrum method, such as its accuracy and ease at which the
reconstruction at various distances can be computed.

Consider imaging a slanted surface, which is several wavelengths high ( Figure 7.7A) . The
corresponding phase map will then exhibit multiple phase jumps ( Figure 7.7B) . We can
select a part of the phase map (B) with values between 0 and Δz ( Figure 7.7C) , which are
far from any discontinuities. We then pick a certain area (marked in (C)) as a “starting
point” and discard all the other areas that are not adjacent to it, as they are likely from the