Detection Using a Generalized Hough Transform Track-Before-Detect Processing

Under this SBIR program, ORINCON proposes to demonstrate algorithms to perform detection, localization, and tracking of targets in shallow-water scenarios where the primary measurements available are bearing (time difference of arrival at two phase centers) versus time. The approach used is Track-Before-Detect processing. The term Track-Before-Detect refers to the joint estimation-detection process involved when several time epochs of the basic measurements are combined in such a way as to include the temporal processing gain available from integrating over the total time span associated with the measurements. This utilization of the past data samples however does not need to imply an equivalent time-late as the processing will be done recursively and the information from the past is stored in the state parameters associated with the hypothesized tracks of the targets. A similar more restrictive process is currently being used at ORINCON in connection with the Full Spectrum Program. The restrictive feature alluded to is that the presently implemented ¿Hough Transform¿ assumes the time difference versus time track is sufficiently short enough to be approximated by a straight line. The utilization of the ¿Generalized Hough Transform¿ as proposed by NRL will remove this restriction.