Browse Prior Art Database

Profile Dependent Piece Correlation Algorithm

IP.com Disclosure Number: IPCOM000103989D
Original Publication Date: 1993-Feb-01
Included in the Prior Art Database: 2005-Mar-18
Document File: 1 page(s) / 37K

Publishing Venue

IBM

Related People

Kennedy, ET: AUTHOR

Abstract

Disclosed is an algorithm for correlating missile boost track segments which have a significant gap between segments. Each track segment consists of two dimensional observations from sensors and may be either monocular or multiple satellite viewed. One of the track segments in a pair must have been previously characterized by a typing algorithm which matches the track segment with a profile, thus obtaining states for that segment. The profile, along with the states, locates the track segment three-dimensionally in space. The states and typed profile are used to compute the predicted positions at the time of the returns of the to be correlated segment. First, the projections of these predictions in the reference frame(s) of the sensor(s) observing the to be correlated segment is computed.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 83% of the total text.

Profile Dependent Piece Correlation Algorithm

      Disclosed is an algorithm for correlating missile boost track
segments which have a significant gap between segments.  Each track
segment consists of two dimensional observations from sensors and may
be either monocular or multiple satellite viewed.  One of the track
segments in a pair must have been previously characterized by a
typing algorithm which matches the track segment with a profile, thus
obtaining states for that segment.  The profile, along with the
states, locates the track segment three-dimensionally in space.
     The states and typed profile are used to compute the predicted
positions at the time of the returns of the to be correlated segment.
First, the projections of these predictions in the reference frame(s)
of the sensor(s) observing the to be correlated segment is computed.
If these projections are close enough to the actual observations,
then additional computations are performed.  The returns from the
two-track segments are then fitted to the typed profile using a
three-state filter.  If the score is sufficiently small, the
two-track segments are considered to be potentially correlated
pending the result of a greedy assignment algorithm.  If the score
fails the test limit, but only marginally, then other profiles which
are similar to the typed profile are fit to the observations from the
two tracks.  If any of these fits are successful, then the track
segments are considered potential...