Abstract:
Polarimetric synthetic aperture radar (PolSAR) is an active imaging radar system that remotely
senses the target/area illuminated by microwaves which is employed in varied range of applications
from agriculture, forest, urban and military. The ability of the system to be operated in
all weather and all time due to it’s operation in a microwave region in the frequency spectrum
has made this system a popular one. PolSAR radar gets the description about the target and
it’s physical properties using polarization property of the electromagnetic (EM) waves. A fully
polarimetric radar simultaneously transmits and receive microwave pulses with two different
orthogonal polarizations. It measures the back scatter coefficients at two orthogonal polarizations,
and forms a image of the target that is being observed. Fully polarimetric measurements
thus constitute a complete description of the scattering characteristics of the target in a single
resolution cell. In radar polarimetry, the complex scattering coefficients measured at different
polarizations are used to characterize the target.
Edge detection in PolSAR images is an important image processing step employed for processing
the PolSAR data. It entails finding the change in intensity level in an image so as to
register it as an edge. This is a basic and necessary processing step for further ”segmentation,
registration, and identification” of the features in PolSAR data [1]. An edge detector finds
discontinuity or change in sudden intensity level in an image, this detected edge may further be
utilized for processing the image to generate various segments based on some defined characteristics
or to map the images to certain physical parameters.
This thesis unfolds in two major section. Firstly it describes various edge detection methods
that are currently being used in field of PolSAR imagery, with their suitability and compatibility
in application to PolSAR data. It also gives on account of detail into necessity of contrast
enhancement in the process of edge detection. There after a novel technique is introduced
comprising of an algorithm that efficiently enhances the PolSAR image contrast by entropy
v
Abstract
maximization and robustly detects the edges. Further an application of edge detection in order
to detects the road and boundaries in a rural area is discussed with data obtained from various
airborne and space borne radars.
