Spectral Image Segmentation for Target Material Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile remote sensing platforms face challenges in identifying target materials due to unknown characteristics of targets and backgrounds, particularly with platforms that cannot be remotely configured in real or near real-time, and hyperspectral imaging increases processing and communication demands without fully addressing sensor resolution and update concerns.
Innovation Solution
A system and method that acquires spectral images, segments them into material areas, and uses spectral models to identify target materials by comparing data with databases, outputting identifiers for display, which includes a processor-based apparatus and computer-readable medium for implementing a target material identification system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hyperspectral imagers are used to collect detailed spectral information, then measurement precision is improved, but device complexity and processing requirements increase
Solution Approach 1:
The patent segments the spectral image into multiple regions based on spectral similarity, allowing processing to be performed on smaller regional subsets rather than the entire image. This reduces computational complexity while maintaining measurement precision through region-specific spectral model matching.
Solution Approach 2:
The patent applies spectral model matching selectively to identified regions rather than processing all pixels uniformly. By focusing computational resources on regions containing potential target materials and using statistical criteria to filter matches, the system achieves high measurement precision without requiring excessive processing power across the entire image.
2Ease of operation
If hyperspectral sensing is used to ease spectral band selection, then ease of operation is improved, but productivity decreases due to sensor resolution and update concerns
Solution Approach 1:
The patent pre-establishes a database of spectral models for various materials before field deployment. During operation, the system quickly matches observed spectral signatures against this pre-prepared database using statistical criteria, eliminating the need for real-time spectral band selection and enabling rapid target identification that maintains high productivity.
3Measurement precision
If spectral images are processed without segmentation, then measurement precision is maintained, but loss of time increases due to processing burdens
Solution Approach 1:
The patent divides the spectral image into multiple regions based on spectral characteristics, allowing parallel processing of each region. This segmentation maintains measurement precision by enabling region-specific spectral model matching while significantly reducing overall processing time through distributed computation and focused analysis of only relevant regions.
Solution Approach 2:
The patent extracts and processes only the spectral regions that contain potential target materials, rather than processing the entire image uniformly. By using statistical criteria to identify and isolate regions of interest, the system maintains high identification accuracy while minimizing processing time by excluding irrelevant areas from detailed analysis.
Data Source
AI summary
A method of identifying a target material in a spectral image includes acquiring a spectral image of a scene. The method also includes performing image segmentation to partition the spectral image into a plurality of segments. The method includes accessing a database of spectral models of a plurality of materials to determine a material whose spectral model is most similar to the spectral data for the segment, a difference between the spectral model of the material and the spectral data for the segment including measurable reflectance or radiance at characteristic frequencies or wavelengths. The method also includes analyzing a database of spectral data for a plurality of target materials to identify a target material whose spectral data also has measurable reflectance or radiance at the characteristic frequencies or wavelengths. And the method includes outputting an identifier of the target material for display with the spectral image.