Hyperspectral Virtual Bands for Single-Pixel Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hyperspectral data processing techniques face challenges in handling large volumes of high-dimensional data and detecting small objects like single-pixel or subpixel anomalies, as dimension reduction methods often make these anomalies undetectable.
Innovation Solution
A system using particle swarm optimization to determine optimal virtual bandwidths for combining hyperspectral data cubes, reducing data volume and enhancing anomaly detection performance by transforming input data into virtual bands with optimized rectangular filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If dimension reduction techniques are used to process hyperspectral data, then the data processing complexity is reduced, but single-pixel and subpixel anomaly detection capability deteriorates
Solution Approach 1:
The patent transforms the hyperspectral data from the spatial domain to the wavelength domain by combining multiple spectral bands into a smaller number of virtual bands with optimized bandwidths. This parameter transformation allows the data to be processed in a different domain where anomaly detection is more effective, resolving the contradiction between processing complexity and detection capability
Solution Approach 2:
Instead of reducing dimensions within the spatial domain, the patent moves the processing to the wavelength domain, creating virtual bands that represent combinations of original spectral bands. This dimensional transformation enables both reduced complexity and maintained detection capability by operating in a more suitable domain for the detection task
2Loss of information
If all wavelength bands are processed, then complete spectral information is retained, but computational load increases significantly
Solution Approach 1:
The patent combines multiple original spectral bands into a smaller number of virtual bands by merging their information in the wavelength domain. This combining process reduces the total number of bands that need to be processed, significantly lowering computational load while the optimization ensures that the most informative combinations are retained
Solution Approach 2:
The patent changes the parameter representation from individual spectral bands to virtual bands with optimized bandwidths. By transforming the data into this new parameter space, the system achieves better signal-to-noise ratio and reduces the dimensionality of the problem, thereby reducing computational requirements without losing critical spectral information
3Measurement precision
If virtual bands with optimized bandwidths are used, then anomaly detection performance is improved, but data transformation complexity increases
Solution Approach 1:
The patent performs preliminary optimization to determine the optimal number and bandwidths of virtual bands before the actual anomaly detection process. By pre-calculating the optimal virtual band parameters using training data or statistical analysis, the system establishes a transformation scheme that maximizes detection performance while avoiding the need for complex real-time optimization during detection
Solution Approach 2:
The patent introduces a wavelength domain transformation that maps original spectral bands to virtual bands with optimized bandwidths. This transformation to another domain simplifies the detection problem by concentrating the spectral information in a more effective representation, improving anomaly detection performance while the transformation itself becomes a manageable preprocessing step
Data Source
AI summary
Described is a system for anomaly detection from hyperspectral data. The system receives, from a hyperspectral sensor, input hyperspectral data cubes with input data bands for every pixel. Using an optimization technique, an optimized number of virtual bands is determined to use for combining the hyperspectral data cubes to achieve optimal anomaly detection performance. The optimized number of virtual bands is less than the input data bands, resulting in reduced hyperspectral data. The hyperspectral data cubes are combined into the optimized number of virtual bands with the optimal bandwidth for each virtual band, and a set of combined hyperspectral data is output. Anomalies are detected in the set of combined hyperspectral data using an anomaly detection technique. Based on the detected anomalies, single-pixel or subpixel targets of interest are detected in the combined hyperspectral data.