Householder PAN Sharpening for Hyperspectral Imagery

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing multi-spectral (MS) image sharpening techniques, such as Gram-Schmidt PS, face numerical instability issues when dealing with a large number of bands, limiting their effectiveness for superspectral and hyperspectral imagery.

Innovation Solution

The Householder PAN Sharpening (HPS) method is employed, which provides numerical stability by using a Householder transformation to orthogonalize a combined matrix of pseudo-PAN and MS image bands, allowing for correct results with a large number of bands, including superspectral and hyperspectral imagery.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If Gram-Schmidt PS is used for MS image sharpening, then the sharpening process can be performed, but numerical instability occurs when dealing with a large number of bands

Engineering Contradiction:

Improvenumerical stabilityVSAvoidapplicability to hyperspectral imagery

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent changes the mathematical parameter/algorithm from Gram-Schmidt orthogonalization to Householder QR orthogonalization. This parameter change in the computational method provides numerical stability for handling large numbers of bands (20+ bands) in hyperspectral imagery, resolving the contradiction between reliability and adaptability.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If MS image sharpening is performed to increase spatial resolution, then image clarity improves, but numerical instability limits the number of bands that can be processed

Engineering Contradiction:

Improvespatial resolutionVSAvoidnumerical stability

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent applies Householder QR orthogonalization instead of Gram-Schmidt PS, changing the computational parameter to achieve both high spatial resolution sharpening and numerical stability. This allows processing of hyperspectral imagery with 20+ bands while maintaining measurement precision.

Inventive Principle:

Principle #35Parameter changes

3Quantity of substance

If the number of spectral bands is increased for superspectral and hyperspectral imagery, then spectral information improves, but existing sharpening techniques become numerically unstable

Engineering Contradiction:

Improvenumber of spectral bandsVSAvoidnumerical stability

Core Design Contradiction:

Quantity of substanceVSReliability

Solution Approach 1:

The patent changes the orthogonalization algorithm parameter from Gram-Schmidt to Householder QR method, enabling stable processing of imagery with a large quantity of spectral bands (20+ bands) in superspectral and hyperspectral applications.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10147170B2Systems and methods for sharpening multi-spectral imagery

Publication Date: 2018.12.04 RAYTHEON CO

AI summary

Discussed herein are apparatuses, systems, and methods for sharpening multi-spectral image data using panchromatic image data. A method can include using a Householder transform in such sharpening.