Compressed Hyperspectral Detection via Basis Vector Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hyperspectral imaging data processing requires decompression of compressed data for target detection, which increases computational intensity and resource burden, as it involves extensive processing steps and vector or matrix operations, making it inefficient for real-time detection in highly compressed forms.
Innovation Solution
A method and system that utilize basis vectors to reduce the spectral reference and compute a detection filter score directly from compressed hyperspectral data, allowing for detection of materials without full-dimensional processing, by forming an M-dimensional spectral reference detection filter from an N-dimensional spectral reference and scene covariance, and comparing it to a threshold to determine the presence of target materials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If compressed hyperspectral data is decompressed for target detection, then detection accuracy is improved, but computational workload and processing time increase significantly
Solution Approach 1:
The patent transforms the detection problem from full spectral dimension to a compressed dimension space using basis vectors. Instead of operating on all spectral bands, the method projects both the scene pixel and spectral reference onto a lower-dimensional subspace spanned by selected basis vectors, performing detection operations in this reduced dimension while maintaining detection accuracy.
Solution Approach 2:
The patent extracts only the essential spectral information needed for detection by selecting M basis vectors from the full spectral dimension N. This extraction process identifies and retains the most relevant spectral components that characterize target materials, discarding redundant information while preserving detection capability.
2Reliability
If full-dimensional spectral processing is performed, then detection reliability is improved, but computational resources and processing burden increase
Solution Approach 1:
The patent changes the spectral representation parameters by expressing scene pixels and spectral references in terms of basis vector coefficients rather than full spectral values. This parameter transformation allows detection operations to be performed on compressed coefficient representations, reducing computational complexity while maintaining detection reliability through proper basis vector selection.
3Productivity
If compressed data is used directly for detection, then processing efficiency is improved, but detection accuracy deteriorates without proper spectral reference adaptation
Solution Approach 1:
The patent performs preliminary adaptation of the spectral reference to the compressed dimension space before detection operations. By pre-computing the compressed spectral reference and its covariance matrix in the basis vector space, the method ensures that detection algorithms operate with properly adapted reference data, maintaining accuracy while enabling efficient compressed-domain processing.
Data Source
AI summary
Provided are examples of a detecting engine for identifying detections in compressed scene pixels. For a given compressed scene pixel having a set of M basis vector coefficients, set of N basis vectors, and code linking the M basis vector coefficients to the N basis vectors, the detecting engine reduces a spectral reference (S) to an N-dimensional spectral reference (SN) based on the set of N basis vectors. The detecting engine computes an N-dimensional spectral reference detection filter (SN*) from SN and the inverse of an N-dimensional scene covariance (CN). The detecting engine forms an M-dimensional spectral reference detection filter (SM*) from SN* based on the compression code and computes a detection filter score based on SM*. The detecting engine compares the score to a threshold and determines, based on the comparison, whether the material of interest is present in the given compressed scene pixel and is a detection.