Data processing: artificial intelligence – Knowledge processing system – Knowledge representation and reasoning technique
Reexamination Certificate
1999-06-30
2002-05-14
Black, Thomas (Department: 2122)
Data processing: artificial intelligence
Knowledge processing system
Knowledge representation and reasoning technique
C382S156000, C382S157000
Reexamination Certificate
active
06389408
ABSTRACT:
GOVERNMENT INTEREST
The invention described herein may be manufactured, licensed, and used by or for the U.S. Government.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a neural network pattern recognition system that identifies chemical and/or biological materials (CBMs) at a distance by recognizing the material's light scattering signature. More particularly, the present invention trains a neural network system to perform pattern recognition from susceptible Mueller matrix elements derived from modulated polarized infrared laser light that is scattered from a contaminated surface on and off an absorption band of the CBM target (contaminant). The neural network acts as a filter that identifies any-of-N compounds in real-time through their unique differential-absorption Mueller matrix properties.
2. Brief Description of the Related Art
Infrared (IR) luminescence, polarized scattering, and volume reflectance technologies have been evaluated as standoff detection methods for chemical and biological warfare agents (CBWA) deposited on terrain and man-made landscapes. Back-reflectance techniques have involved spectroscopic measurements of a depolarized, multiple-scattered subsurface IR radiance component from soil and sand wetted by several simulants of liquid chemical warfare agents. Generally these methods have proven insensitive as they tend to detect a simulant's absorption bands far above a threshold volume concentration considered life threatening had the contaminant been nerve agent, such as VX. However, the active multiwavelength polarized backscattering and “pseudo-active” thermal luminescence methods of remote detection have shown improved sensitivity and accuracy. One active technique called differential absorption Mueller matrix spectroscopy, or DIAMMS, has shown promise for solving the combined CBWA detection problem.
Neural network analysis of polarized light scattering can be applied to identify biological and chemical contaminants by a system in which monochromatic polarization-modulated infrared laser light is backscattered by the contaminants at two close wavelengths. One beam wavelength corresponds to an absorption band of the scatterer while the other wavelength is nonabsorbing. The light beams backscattered by the contaminant are examined in the form of a Mueller matrix whose elements describe all polarization states of the backscattered light at each pixel of the detector array. Mueller matrices are mathematical calculations and representations of irradiated materials. They consist of 16 elements, and completely describe both the geometric (particle size) and physical (refractive index) aspects of the scatterer. These neural network systems filter the unique pattern of susceptible Mueller matrix elements of 15 or less normalized difference-elements of the CBMs. Characteristics such as particle size, particle shape, refractive index, and the like, can be correlated.
U.S. Pat. No. 4,306,809 (Azzam), U.S. Pat. No. 4,953,980 (DeVolk et al.), U.S. Pat. No. 4,884,886 (Salzman et al.), U.S. Pat. No. 5,247,176 (Goldstein), U.S. Pat. No. 5,631,469 (Carrieri et al.), U.S. Pat. No. 5,659,391 (Carrieri), and U.S. Pat. No. 5,708,503 (Carrieri) disclose many aspects of passive and active systems which process and transform scattergrams into Mueller elements, the disclosures of these patents are incorporated herein by reference.
SUMMARY OF THE INVENTION
In view of the foregoing, it is an object of the present invention to provide neural network training sufficient for CBWA detection.
It is further an object of the present invention to provide training of neural networks for selective CBM identification and improved efficiency of using Mueller matrix elements in CBM identification.
These and other objects are accomplished with the present invention that includes a neural network pattern recognition system for remotely sensing and identifying chemical and biological materials comprising a software component having an adaptive gradient descent training algorithm capable of performing backward-error-propagation and an input layer is formatted to accept differential absorption Mueller matrix spectroscopic data, a filtering weight matrix component capable of filtering pattern recognition from Mueller data for specific predetermined materials, and, a processing component capable of receiving the pattern recognition from the filtering weight matrix component and determining the presence of specific predetermined materials.
The present invention further includes a method for sensing and identifying chemical and biological materials comprising a software component having an adaptive gradient descent training algorithm capable of performing backward-error-propagation and an input layer is formatted to accept differential absorption Mueller matrix spectroscopic data, a filtering weight matrix component capable of filtering pattern recognition from Mueller data for specific predetermined materials, and, a processing component capable of receiving the pattern recognition from the filtering weight matrix component and determining the presence of specific predetermined materials, building artificial neural network systems for detecting specific solid organic compounds by pattern recognition of their polarized light scattering signatures, and discerning the presence of specific analytes within a sample based upon cued susceptive Mueller matrix difference elements.
REFERENCES:
patent: 5631469 (1997-05-01), Carrieri et al.
Biffoni Ulysses John
Black Thomas
Holmes Michael B.
Ranucci Vincent J.
The United States of America as represented by the Secretary of
LandOfFree
Neural network systems for chemical and biological pattern... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Neural network systems for chemical and biological pattern..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Neural network systems for chemical and biological pattern... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2877050