System and method for analyzing optical spectral data of a terrain image

A system for analyzing data from an image, said system comprising:

• spectrometer for receiving portions of said image and producing a plurality of spectra of individual pixels in said image;
• light sensitive detector for receiving said plurality of spectra and converting said spectra into electrical data signals;
• and
• target recognition unit for processing said electrical data signals representing said spectra, said target recognition unit including correlation means for comparing said signals with previously stored spectra data from features of known classes, said correlation means generating a correlation measure of the similarity between said spectra for said pixels and said spectra for said known feature, said correlation means including means for storing a matrix of coefficients, each of said coefficients in said matrix corresponding to one of the wavelength values in said spectra for one of a plurality of spectral basis vectors corresponding to known feature classes;
• said correlation means further including multiply and accumulate means for receiving said coefficients, from said means for storing as well as said electrical data signals, said multiply and accumulate means multiplying said spectral data signals by a corresponding coefficient in the matrix, and the multiply and accumulate means adding the resulting multiplicative products to generate a spectral measure for each pixel in a image, wherein the coefficients in the matrix are chosen such that the results of the multiplication yield both positive and negative values and the sum will be approximately zero where there is no match and will be approximately one where there is a match and wherein a plurality of resulting sums for each pixel is a correlation value which indicates the likelihood that the spectra for the pixel is from one of the spectral basis vectors.

A mobile image gathering system for transmitting data from an image to a receiving unit, said system comprising:

• a spectrometer for receiving portions of said image and producing a plurality of spectra of individual pixels in said image;
• a light sensitive detector for receiving said plurality of spectra and for converting said spectra into the electrical data signals;
• a target recognition unit for processing said electrical data signals representing said spectra, said target recognition unit including correlation means for comparing said signals with previously stored spectra data from known classes of features, said correlation means generating a correlation measure of the similarity between said spectra for said pixels and said spectra for said known features, said correlation means including means for storing a matrix of coefficients, each of said coefficients in said matrix corresponding to one of the wavelength values in said spectra for one of a plurality of spectral basis vectors corresponding to known feature classes;
• said correlation means further including multiply and accumulate means for receiving said coefficients, from said means for storing as well as said electrical data signals, said multiply and accumulate means multiplying said spectral data signals by a corresponding coefficient in the matrix, the multiply and accumulate means adding the resulting multiplicative products to generate a spectral measure for each pixel in a image, wherein the coefficients in the matrix are chosen such that the results of the multiplication yield both positive and negative values and the sum will be approximately zero where there is no match and will be approximately one where there is a match and wherein a plurality of resulting sums for each pixel is a correlation value which indicates the likelihood that the spectra for the pixel is from one of the spectral basis vectors;
• and
• means for transmitting said correlation measure to said receiving unit, wherein said correlation measure is transferred to said receiving unit.

A method for processing image data for transmission from a vehicle to a receiving unit, said method comprising:

• receiving a plurality of pixels of an image in a spectrometer;
• generating a spectra for each said pixel;
• converting said spectra into electrical data signals;
• comparing said spectra data with previously stored spectra data from known classes of features;
• by storing a matrix of coefficients, each of said coefficients in said matrix corresponding to one of the wavelength values in the spectra for one of a plurality of spectral basis vectors corresponding to known feature classes, multiplying the spectral data signals by a corresponding coefficient in the matrix, and adding the resulting multiplicative products to generate a spectral measure for each pixel in an image, wherein the coefficients in the matrix are chosen such that the results of multiplication yield both negative and positive values and the sum will be approximately zero where there is no match and will be approximately one where there is a match;
• generating a correlation measure for said pixels from said spectral measure which indicates the likelihood that the spectra for the pixel is from one of the spectral basis vectors;
• and
• transmitting said correlation measure for each pixel from said vehicle to said receiving unit.