Spectral Image Relationship Extraction for Subpixel Target Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-spectral imaging technologies face challenges in detecting and classifying subpixel targets, especially in low spectral contrast environments, due to degraded performance of traditional methods when using pure 100% fill-fraction spectral libraries, particularly in Long Wave Infra-Red (LWIR) imagery where target and background spectra are poorly separated.
Innovation Solution
The Spectral Image Relationship Extraction (SPIRE) method, which employs spectral and spatial relationships for subpixel target detection and classification, uses local and global pixel relationships, localized background estimation, and spectral demixing techniques to identify targets in real-time, even at low fill-fractions and in challenging spectral conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional relative target to background thresholding methods are used after best match determination against a target spectral library, then detection and classification can be performed in multi-spectral images, but performance degrades significantly when detecting subpixel targets in low spectral contrast environments with pure 100% fill-fraction target spectral libraries
Solution Approach 1:
The patent transforms the detection approach by changing from direct spectral matching against pure 100% fill-fraction libraries to an unmixing-based approach that separates target and background spectral components. This parameter change in the detection methodology enables reliable subpixel target detection in low spectral contrast environments where traditional methods fail.
Solution Approach 2:
The patent introduces an intermediary unmixing process that acts as a mediator between the raw multi-spectral image data and the target spectral library. This intermediary step decomposes mixed pixel spectra into pure target and background components, enabling accurate detection of subpixel targets without requiring modified spectral libraries.
2Reliability
If spectral demixing and local global reconciliation are applied to identify subpixel targets, then detection rate increases by 20-50% and false alarms reduce by 50-70%, but computational complexity increases
Solution Approach 1:
The patent segments the detection process into distinct modular stages: spectral demixing to separate target and background components, candidate identification based on spectral residuals, and local-global reconciliation to confirm detections. This segmentation improves detection reliability while managing computational complexity through structured processing steps.
Solution Approach 2:
The patent performs preliminary spectral demixing and background subtraction before target identification, preparing the data in advance to facilitate more accurate detection. This preliminary action separates the target spectral signature from background clutter, making subsequent detection more reliable and reducing false alarms.
3Ease of manufacture
If pure 100% fill-fraction target spectral libraries are used for subpixel target detection, then the spectral library remains simple and standardized, but target and background matching results are no longer cleanly separated causing method degradation
Solution Approach 1:
The patent extracts the target spectral component from mixed pixel spectra through spectral unmixing operations. By mathematically separating the target contribution from the background contribution, the system maintains use of simple pure spectral libraries while achieving accurate target-background separation for subpixel detection.
Solution Approach 2:
The patent replaces the mechanical approach of using physically separated target and background pixels with a mathematical unmixing approach. Instead of requiring physically distinct spectral signatures, the system uses spectral decomposition algorithms to separate mixed signals, maintaining library simplicity while achieving precise separation accuracy.
Data Source
AI summary
Real-time subpixel detection and classification is provided. The method comprises receiving input of a spectral library of targets, a multi-spectral image cube, and a list of background image samples. A candidate ground spatial distance (GSD) cell within the multi-spectral image cube is selected for spectral demixing. The spectrally demixed candidate GSD cell is compared against the spectral library and the list of background image samples. A determination is made whether the candidate GSD cell contains an identifiable target. The candidate GSD cell is labeled unknown if it does not resemble a target in the spectral library nor a sample in the list of potential background image sample. Local global reconciliation is applied to the candidate GSD cell to reject false detections of non-targets and confirm true detection of targets. Detected targets from the candidate GSD cell or an unknown are output in real-time.