System and method for discriminating constituents of image by complex spectral signature extraction

A system for discriminating constituents of an image electronically captured with complex spectral content by automatic spectral signature extraction comprising:

• a general dictionary having a plurality of reference components;
• a discovery processor coupled to said general dictionary, the discovery processor being programmably configured to execute a simultaneous sparse approximation process upon a set of image data pixels for at least a portion of the captured image, the discovery processor further programmably configured to adaptively discover for the image data vectors a set of core spectral signatures defined in terms of reference components automatically selected from said general dictionary;
• a spectral signature dictionary coupled to said discovery processor for storing said core spectral signatures discovered thereby;
• and,
• a detection processor programmably configured to execute a decomposition process for decomposing individual captured image data pixels to a representation as a linear combination of said core spectral signatures;
• whereby constituent features of the captured image portion corresponding to said linearly combined core spectral signatures are detected in the decomposed image data pixel.

A method for processing electronically captured images formed by a plurality of spectral components comprising the steps of:

• electronically capturing image data with complex spectral content, the image data including a plurality of image data vectors defined with respect to a spatial image plane;
• establishing a plurality of candidate spectral components;
• actuating a discovery processor to execute a simultaneous sparse approximation process upon image data vectors of at least a portion of the captured image, actuating the discovery processor to adaptively discover for the image data vectors a set of core spectral signatures defined in terms of candidate spectral components automatically selected from a general dictionary;
• and,
• selectively matching at least one individual captured image data vector to at least a portion of said core spectral signature set to decompose said image data vector to a representation as a linear combination of the portion of said core spectral signature set;
• whereby constituent features of the captured image portion corresponding to said selectively matched core spectral signatures are discriminated according to said linear combination thereof.