A Method for Tracking Targets with Hyper-Spectral Data
Abstract
It is a general purpose and object of the present invention to provide a method for tracking both the spatial sensor data and hyper-spectral sensor data associated with a target. It is a further purpose to estimate a frequency spectrum for a target contribution that consists of only the target's energy contribution as opposed to the target energy and the noise energy together. These objects are accomplished with the present invention by taking the histogram model used in H-PMHT and extending it to treat the problem of tracking using hyper-spectral data. In the present invention each measurement scan is now a multi- dimensional array wherein each spatial cell has an associated vector of amplitudes in several (possibly disjoint) spectral cells. The intensity data in the multi-dimensional array is interpreted as the spatial-spectral histogram of a synthetic shot process. A statistical model of the random variation of individual cell intensities from scan to scan is required. The procedure adopted in H-PMHT is to quantize the data vector into a "pseudo-histogram," and then use a multinomial distribution to model the cell counts where the PMHT target mixture model parameterizes the multinomial distribution. The target mixture model determines the cell probabilities that correspond to expected cell counts.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 07, 2003
- Accession Number
- ADD020111
Entities
People
- Roy L. Streit
Organizations
- United States Department of the Navy