Abstract:
Correlation techniques with number of variations have been adopted very effectively
in various scenarios of object detection in the past. An important breakthrough
achieved in correlation field is the development of the filter which displays improved
performance in clutter rejection, noise suppression and out of plane rotation
invariance.
“Extended maximum average correlation height” is the latest development
encompassing all the desired capabilities in a correlation filter. It displays remarkable
results which allow usage of this filter in military based applications handling real
time data.
Although EMACH filter has been well researched and mathematically modeled by
Mohammad Alkanhal and B.V.K Vijaya Kumara in a “SPIE” research paper titled as
“Removing False Alarms in MACH Filter”, but it lacks effective and speedy
implementation being computationally intensive. Moreover, training of filter for
multiple images turns out into non trivial calculations resulting in a heavy workload
for personal/ commercial computers.
