Hyperspectral target identification

Patent metadata

Field	Value
Publication number	US-2021241014-A1
Application number	US-202016782108-A
Country	US
Kind code	A1
Filing date	Feb 5, 2020
Priority date	Feb 5, 2020
Publication date	Aug 5, 2021
Grant date	—

Abstract

The system and method for hyperspectral target identification provides for sue of a sensor array such that for a 2 by 2 pixel, each pixel is set to a different wavelength by employing a band pass filter, so that one can determine whether a cluster set is of a natural object or a man-made object. Using ratios of the collected energy within the partitioned sub-bands one can make near real-time declarations about targets and in some cases, friend or foe determinations.

First claim

What is claimed:

1 . A method for hyperspectral target identification, comprising:
mapping specific wavelengths and bandwidths to each of a plurality of pixels in a cluster within a sensor array using a series optical filters for the specific clutter signatures; calculating ratios between pixel pairs from the plurality of pixels on the sensor array, wherein at least four pixels are used for calculating ratios and one of the four pixels is used as a reference pixel; and eliminating clutter from an image comprising the plurality of pixels on the sensor array based on the calculated ratios using a programmable logic device having contrast spatial filters.

2 . The method for hyperspectral target identification according to claim 1 , wherein the programmable logic device is an FPGA.

3 . The method for hyperspectral target identification according to claim 1 , wherein the elimination of clutter includes the removal of vegetation from the image.

4 . The method for hyperspectral target identification according to claim 1 , wherein the elimination of clutter from the image further comprises using confidence levels to determine whether an object is vegetation, water, soil, or some other object.

5 . The method for hyperspectral target identification according to claim 1 , further comprising determining whether an object in the image is friend or foe.

6 . The method for hyperspectral target identification according to claim 1 , wherein the at least four pixels is a 2×2 pixel grouping a 1 st pixel is full band of the sensor, a 2 nd pixel is 1.05 μm, a 3 rd pixel is 1.50 μm, and a 4 th pixel at 1.95 μm, wherein the wavelengths are defined by the spectral response of the target and the clutter to be removed.

7 . The method for hyperspectral target identification according to claim 1 , wherein the method continues by shifting a single pixel and repeating the process, pixel by pixel, row by row across the plurality of pixels on a sensor array.

8 . The method for hyperspectral target identification according to claim 1 , further comprising using automatic target recognition (ATR) to differentiate between target types based on a priority list.

9 . The method for hyperspectral target identification according to claim 1 , further comprising using a fusion approach comprising additional sensor data and post processing.

10 . A method for hyperspectral target identification, comprising:
mapping specific wavelengths and bandwidths to each of a plurality of pixels on an FPGA sensor array; calculating ratios between pixel pairs from the plurality of pixels on an FPGA sensor array, wherein at least four pixels (2×2) are used for calculating ratios and one of the four pixels is used as a reference pixel; and eliminating clutter from an image comprising the plurality of pixels on an FPGA sensor array based on the calculated ratios using contrast spatial filters to provide real-time processing.

11 . The method for hyperspectral target identification according to claim 10 , wherein the elimination of clutter include the removal of vegetation from the image.

12 . The method for hyperspectral target identification according to claim 10 , wherein the elimination of clutter from the image further comprises using confidence levels to determine whether an object is vegetation, water, soil, or some other object.

13 . The method for hyperspectral target identification according to claim 10 , further comprising determining whether an object in the image is friend or foe.

14 . The method for hyperspectral target identification according to claim 10 , wherein in a 2×2 pixel grouping a 1 st pixel is full band of the sensor, a 2 nd pixel is 1.05 μm, a 3 rd pixel is 1.50 μm, and a 4 th pixel at 1.95 μm, wherein the wavelengths are defined by the spectral response of the target and the clutter to be removed.

15 . The method for hyperspectral target identification according to claim 10 , wherein the method continues by shifting a single pixel and repeating the process, pixel by pixel, row by row across the plurality of pixels on an FPGA sensor array.

16 . The method for hyperspectral target identification according to claim 10 , further comprising using automatic target recognition (ATR) to differentiate between target types based on a priority list.

17 . The method for hyperspectral target identification according to claim 10 , further comprising using a fusion approach comprising additional sensor data and post processing.

18 . A computer program product comprising one or more non-transitory machine-readable mediums encoded with instructions that when executed by one or more processors cause a process to be carried out for hyperspectral target identification, the process comprising:
mapping specific wavelengths and bandwidths to each of a plurality of pixels on an FPGA sensor array; calculating ratios between pixel pairs from the plurality of pixels on an FPGA sensor array, wherein at least four pixels (2×2) are used for calculating ratios and one of the four pixels is used as a reference pixel; and eliminating clutter from an image comprising the plurality of pixels on an FPGA sensor array based on the calculated ratios using contrast spatial filters to provide real-time processing.

Assignees

• Bae Sys Inf & Elect Sys Integ

Inventors

• Choiniere Michael J

Classifications

• G06V10/80

  Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level (multimodal speaker identification or verification G10L17/10) · CPC title

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/25

  Fusion techniques · CPC title

• G06F18/2431

  Multiple classes · CPC title