Hyperspectral remote sensing systems and methods using covariance equalization

Abstract

A method and apparatus for detecting a target or targets in a surrounding background locale based on target signatures obtained by a hyperspectral imaging sensor used the hyperspectral imaging sensor to collect raw target signature data and background locale data during a first data collection mission. The data is processed to generate a database including a plurality of target signatures and background data relating to the background locale. The hyperspectral imaging sensor is later used to collect further background data during a further, current data collecting mission so as to provide continuously updated background data, in real time. A covariance equalization algorithm is implemented with respect to the background data contained in the database and the updated background data collected during the current mission to effect transformation of each target signature of the database into a transformed target signature. A detection algorithm which employs the resultant transformed target signature is used to produce detection information related to the target or targets.

Description

FIELD OF THE INVENTION [0001] The present invention relates to hyperspectral detecting or sensing systems for detecting military and intelligence targets for other uses. BACKGROUND OF THE INVENTION [0002] The detection of military and intelligence targets by airborne autonomous electro-optical reconnaissance systems is an important evolving capability associated with the maturation of hyperspectral (HS) camera technology. In recent years it has become clear that the primary factor limiting the effectiveness of these systems is no longer hardware performance, but rather the absence of useful knowledge about the in situ spectra of the targets. A similar problem exists in commercial applications. Laboratory reflectance spectra of materials sought by remote sensing systems must be translated into the spectral intensities measured in the field with an imaging spectrometer. [0003] The HS detection algorithms that have been exercised by numerous researchers over the past decade basically f...

AI Technical Summary

Benefits of technology

[0017] An important feature of the invention is that it drastically reduces the performance-degrading effects of dynamic variability in hyperspectral sensing systems. The invention does this by transforming target spectral signatures derived from one image to those appropriate to a second image thereby compensating for dynamic variability. This approach greatly enhances the detectability of similar targets at other locales in an image, and it facilitates an enhanced method of change detection. In the latter application, the transformation provided allows the assessment of change to be based not merely on a comparison of pixels deemed anomalous in each of their respective scenes, but also be based on a test of target similarity.

Problems solved by technology

However, CE requires no detailed models of either the environment or the sensing system.

Images