Showing papers by "Irving S. Reed published in 1997"

Automatic target detection and recognition in multiband imagery: a unified ML detection and estimation approach

Abstract: Multispectral or hyperspectral sensors can facilitate automatic target detection and recognition in clutter since natural clutter from vegetation is characterized by a grey body, and man-made objects, compared with blackbody radiators, emit radiation more strongly at some wavelengths. Various types of data fusion of the spectral-spatial features contained in multiband imagery developed for detecting and recognizing low-contrast targets in clutter appear to have a common framework. A generalized hypothesis test on the observed data is formulated by partitioning the received bands into two groups. In one group, targets exhibit substantial coloring in their signatures but behave either like grey bodies or emit negligible radiant energy in the other group. This general observation about the data generalizes the data models used previously. A unified framework for these problems, which utilizes a maximum likelihood ratio approach to detection, is presented. Within this framework, a performance evaluation and a comparison of the various types of multiband detectors are conducted by finding the gain of the SNR needed for detection as well as the gain required for separability between the target classes used for recognition. Certain multiband detectors become special cases in this framework. The incremental gains in SNR and separability obtained by using what are called target-feature bands plus clutter-reference bands are studied. Certain essential parameters are defined that effect the gains in SNR and target separability.

Subspace selection for partially adaptive sensor array processing

Abstract: This paper introduces a cross-spectral metric for subspace selection and rank reduction in partially adaptive minimum variance array processing. The counter-intuitive result that it is suboptimal to perform rank reduction via the selection of the subspace formed by the principal eigenvectors of the array covariance matrix is demonstrated. A cross-spectral metric is shown to be the optimal criterion for reduced-rank Wiener filtering.

Generalized linear feature detection of weak targets in spectrally mixed clutter

Abstract: The ability to detect weak targets of low contrast or signal-to- noise ratio (SNR) is improved by a fusion of data in space and wavelength from multispectral/hyperspectral sensors. It has been demonstrated previously that the correlation of the clutter between multiband thermal infrared images plays an important role in allowing the data collected in one spectral band to be used to cancel the background clutter in another spectral band, resulting in increased SNR. However, the correlation between bands is reduced when the spectrum observed in each pixel is derived from a mixture of several different materials, each with its own spectral characteristics. In order to handle the identification of objects in this complex (mixed) clutter, a class of algorithms have been developed that model the pixels as a linear combination of pure substances and then unmix the spectra to identify the pixel constituents. In this paper a linear unmixing algorithm is incorporated with a statistical hypothesis test for detecting a known target spectral feature that obeys a linear mixing model in a mixture of background noise. The generalized linear feature detector utilizes a maximum likelihood ratio approach to detect and estimate the presence and concentration of one or more specific objects. A performance evaluation of the linear unmixing and maximum likelihood detector is shown by comparing the results to the spectral anomaly detection algorithm previously developed by Reed and Yu.