Abstract
A phaseless Gauss-Newton inversion (GNI) algorithm is developed for
microwave imaging applications. In contrast to full-data microwave
imaging inversion that uses complex (magnitude and phase) scattered
field data, the proposed phaseless GNI algorithm inverts phaseless
(magnitude-only) total field data. This phaseless Gauss-Newton inversion
(PGNI) algorithm is augmented with three different forms of
regularization, originally developed for complex GNI. First, we use the
standard weighted L2 norm total variation multiplicative regularizer
which is appropriate when there is no prior information about the object
being imaged. We then use two other forms of regularization operators to
incorporate prior information about the object being imaged into the
PGNI algorithm. The first one, herein referred to as SL-PGNI,
incorporates prior information about the expected relative complex
permittivity values of the object of interest. The other, referred to as
SP-PGNI, incorporates spatial priors (structural information) about the
objects being imaged. The use of prior information aims to compensate
for the lack of total field phase data. The PGNI, SL-PGNI, and SP-PGNI
inversion algorithms are then tested against synthetic and experimental
phaseless total field data.