# US 2014/0193078 A1 · Optimized orthonormal system and method for reducing dimensionality of hyperspectral images

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0193078 A1
- Document ID: US20140193078A1
- Status: Published application
- Kind: A1
- Application number: 14/090,753
- Filed: 2013-11-26
- Published: 2014-07-10
- Earliest claimed: 2011-04-13
- Assignees: Raytheon Company
- Applicants: Raytheon Company
- Inventors: Ian S. Robinson; Bradley Flanders
- Web page: https://patentgrep.com/patent/US20140193078A1

## CPC

- G01J 3/0264: {Electrical interface; User interface}
- G01J 3/027: {Control of working procedures of a spectrometer; Failure detection; Bandwidth calculation}
- G01J 3/06: Scanning arrangements {arrangements for order-selection}
- G01J 3/2823: {Imaging spectrometer}
- G06V 20/13: Satellite images
- G01J 2003/2836: {Programming unit, i.e. source and date processing}
- G01J: MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY (light sources F21, H01J, H01K, H05B; investigating properties of materials by optical means G01N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method for reducing dimensionality of hyperspectral images includes receiving a hyperspectral image having a plurality of pixels. The method may further include establishing an orthonormal basis vector set comprising a plurality of mutually orthogonal normalized members. Each of the mutually orthogonal normalized members may be associated with one of the plurality of pixels of the hyperspectral image. The method may further include decomposing the hyperspectral image into a reduced dimensionality image, utilizing calculations performed while establishing said orthonormal basis vector set. A system configured to perform the method may also be provided.

## Claims

### Claim at position 1

1-12. (canceled)

### Claim 13

A method of hyperspectral image dimension reduction, comprising: receiving a hyperspectral image having a plurality of pixels; iteratively establishing a basis vector (BV) set comprising an initial basis vector and individual spectral vectors from the plurality of pixels that have a residual magnitude over a threshold value when a spectral vector of a current iteration is unmixed by unconstrained unmixing with a prior iteration of the BV set; for each of the plurality of pixels: reading a spectral vector from the pixel; and decomposing, by unconstrained unmixing, the spectral vector with the established BV set to provide a reduced dimension vector for the pixel; and providing a dimension reduced image comprising the reduced dimension vectors for each pixel.

### Claim 14 (depends on 13)

The method of claim 13, wherein the residual vector is orthogonal to a prior iteration of the BV set.

### Claim 15 (depends on 13)

The method of claim 13, wherein said basis vector set comprises at least one non-endmernber.

### Claim 16 (depends on 13)

The method of claim 13, wherein said establishing the BV set is configured to minimize error associated with each of the plurality of pixels.

### Claim 17 (depends on 13)

The method of claim 13, wherein the dimension reduction is performed as a lossless process.

### Claim 18 (depends on 13)

The method of claim 13, wherein the dimension reduced image includes an alpha vector and a residual vector for each pixel.

### Claim 19 (depends on 13)

The method of claim 13, wherein establishing a basis vector (BV) set includes:
selecting an initial pixel;
reading the spectral vector for the initial pixel, the spectral vector set assigned as an initial basis vector (BV) of the set; and
for each of the remaining plurality of pixels:
reading a spectral vector from the pixel;
unmixing the spectral vector with the BV set to determine a residual vector; and
in response to a magnitude of the residual vector being greater than a threshold, adding the spectral vector to the BV set.

### Claim 20 (depends on 19)

The method of claim 19, wherein the BV set has a pre-determined maximum number of members, and in response to an attempt to exceed the maximum number, optimizing the set to maintain the maximum number of spectral vectors.