Detection of targets from hyperspectral imagery

摘要

Provided is a process and system for detection of sparse or otherwise weak targets in a hyperspectral image. A hyperspectral image is received having a multitude of pixels, with each pixel having a respective spectrum. In some embodiments, multiple mean spectra are selectively determined for respective sub-regions of the hyperspectral image. The subset mean spectra can be selectively removed from respective pixels, thereby improving image fidelity due to sensor artifacts. Additionally, target detection of such an adjusted image can be determined by one or more of matched filter techniques or by partial un-mixing. In at least some embodiments target detection is enhanced by combining a measure of target match with a residual spectrum determined as a measure of un-match. Target detection can be further improved by application of rules, for example, related to target detection threshold.

主权项

1. A method of determining the presence of a target substance, the method comprising:
receiving a hyperspectral image comprising spectra of a plurality of scene pixels indicative of light and other radiation absorbed by or emitted from, a target substance; determining from a subset of scene pixels a respective mean spectrum, wherein the mean spectrum is an n-dimensional vector of values corresponding to the subset of scene pixels, n being a number of frequency bins; adjusting a spectrum of each scene pixel in the subset through subtraction of the respective mean spectrum from the respective spectrum of the scene pixel; repeating the acts of determining a respective mean spectrum and adjusting each scene pixel for other subsets of scene pixels of the received hyperspectral image; determining for each adjusted scene pixel, a measure of spectral match between a respective spectrum of the scene pixel and a reference spectrum of the target substance; determining for each adjusted scene pixel, a quantitative measure of spectral mismatch between the respective spectrum of the scene pixel and the reference spectrum of the target substance; and detecting a presence of the target substance within the hyperspectral image in response to the determined measure of spectral match and the determined quantitative measure of spectral mismatch, wherein the detecting includes forming a ratio of the measure of spectral match based on mixture coefficients and the measure of spectral mismatch based on a residual spectrum, the ratio being indicative of presence of the target substance within the scene pixel.