(HTML Abstract of)"Dynamic Programming Algorithm for Detecting Dim
Moving Targets" Chapter 4 of book "Multitarget-Multisensor Tracking:
Advanced Applications", Edited by Yaakov Bar-Shalom, Artech House 90.
The subject of this chapter is an innovative track-before-detect method
which uses the dynamic programming algorithm (DPA) to detect/track trajectories
of dim moving targets as registered in an imagery set. The algorithm is very
general and can be applied to different kinds of targets and sensors. However,
to make the discussion more specific, we chose the following particular
example.
A mosaic sensor is staring down from a satellite to a fixed
point on the ground while collecting frames that contain target signatures and
background clutter. The problem is to optimally detect dim moving
targets that cross the sensor's field of view. In this chapter we describe the
algorithm as applied to this case, analyze its performance, and compare
it to conventional algorithms which are based on assembling of thresholded
frames. The DPA's SNR performance is shown to be at least an order of
magnitude higher.