Hyperspectral Image Dimensionality Reduction via Error-Based Basis Vector Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hyperspectral image data sets are excessively large, leading to latency issues during processing and transmission, and existing compression techniques either lose valuable information or fail to provide significant compression, making it difficult to efficiently reduce dimensionality while maintaining relevant data integrity.
Innovation Solution
A method that uses a processor to calculate errors and reduction factors based on basis vectors and spectral dimensions, allowing for selective reduction of hyperspectral image data while maintaining a specified error level or data size, thereby optimizing dimensionality reduction and data volume.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If lossy compression techniques are applied to hyperspectral data sets, then data size is reduced, but valuable information is lost
Solution Approach 1:
The method segments the hyperspectral data processing into two distinct paths: a reduced-resolution path that applies aggressive compression for visualization, and a full-resolution path that preserves complete data for analysis. This segmentation allows different portions of the data to be treated differently based on their intended use, resolving the contradiction between compression and information preservation.
Solution Approach 2:
The patent applies local quality by providing different quality levels to different parts of the data system. The reduced-resolution data stream undergoes significant compression suitable for display, while the full-resolution data stream maintains complete information for computational analysis. Each data path is optimized for its specific purpose, allowing both compression and information retention to coexist in different locales of the system.
2Loss of information
If lossless compression algorithms are used on hyperspectral images, then information is preserved, but significant compression is not achieved
Solution Approach 1:
The system segments the data handling into two parallel streams: one optimized for compression (reduced-resolution) and one for information preservation (full-resolution). This allows the lossless path to maintain integrity without being constrained by the need for high compression ratios, while the lossy path handles the compression burden separately.
Solution Approach 2:
The patent resolves the compression limitation by adding a spatial dimension reduction to the spectral data. Instead of only compressing in the spectral domain, the system creates a downscaled spatial representation, effectively trading spatial resolution for compression ratio in one stream while preserving full resolution in another stream for analysis.
3Productivity
If hyperspectral data is transmitted to remote locations for processing, then processing capability is improved, but transmission time increases
Solution Approach 1:
The method segments the data into two transmission streams: a compressed reduced-resolution stream for rapid transmission and preliminary processing, and a full-resolution stream for detailed analysis. This segmentation enables parallel processing at remote locations, with the compressed stream providing quick initial results while the full-resolution stream undergoes more thorough analysis.
Solution Approach 2:
The patent applies preliminary action by performing initial processing and dimensionality reduction on the hyperspectral data before transmission. The reduced-resolution data undergoes preprocessing that extracts key features and reduces data volume, allowing faster transmission and enabling remote systems to begin analysis sooner while full-resolution data is being transmitted.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods for reducing dimensionality of hyperspectral image data having a number of spatial pixels, each associated with a number of spectral dimensions, include receiving sets of coefficients associated with each pixel of the hyperspectral image data, a set of basis vectors utilized to generate the sets of coefficients, and either a maximum error value or a maximum data size. The methods also include calculating, using a processor, a first set of errors for each pixel associated with the set of basis vectors, and one or more additional sets of errors for each pixel associated with one or more subsets of the set of basis vectors. Utilizing such errors calculations, an optimum size of the set of basis vectors may be ascertained, allowing for either a minimum amount of error within the maximum data size, or a minimum data size within the maximum error value.