Independent covariance estimation and decontamination

摘要

Methods and systems are provided for estimating background spectral content in a hyperspectral imaging (HSI) scene. A HSI processor computes a scene covariance matrix for each of a plurality of sparsely sampled pixel sets, identifies and removes the spectral content of contaminating pixels from the covariance matrices, and checks the consistency among the plurality of decontaminated covariance matrices, iteratively re-sampling and re-computing said matrices until an acceptable consistency is achieved, and then computes a final decontaminated covariance matrix representative of the background spectral content of the scene. Alternate approaches to pixel sampling, and/or using fewer spectral dimensions than are available for the pixels are presented.

主权项

1. An improved method of estimating background spectral content in a hyperspectral scene, the method comprising:
in a covariance estimator of a hyperspectral image processor provided with a hyperspectral scene, acquired by a hyperspectral sensor, comprising spectra of a plurality P of scene pixels indicative of radiation absorbed by, or emitted from, scene materials in a plurality D of spectral dimensions:
sampling N pixel sets each comprising a small number of scene pixels;computing a scene covariance matrix for each of the N pixel sets, each scene covariance matrix representing the variance and correlation of the spectral content of the associated pixel set from the mean of the spectral content of the hyperspectral scene for that pixel set;identifying contaminating pixels from among the sampled pixels using a number M of spectral dimensions from the plurality D of spectral dimensions;decontaminating from each of the covariance matrices the spectral contribution of the contaminating pixels, thereby deriving an associated plurality of decontaminated covariance matrices;checking whether the consistency of the plurality of decontaminated covariance matrices is acceptable;if the consistency of the plurality of decontaminated spectral covariance matrices is acceptable, computing a final decontaminated spectral covariance matrix by averaging the plurality of decontaminated spectral covariance matrices; andif the consistency of the plurality of decontaminated spectral covariance matrices is unacceptable, iteratively repeating the sampling, computing, identifying, decontaminating, and consistency checking steps for a greater number of scene pixels.