Hyperspectral Image Reconstruction via Compressed Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hyperspectral video camera systems face limitations in capturing a large number of wavelength bands due to hardware constraints, leading to inadequate reconstruction accuracy with simple interpolation methods.
Innovation Solution
A compressed sensing framework is employed to reconstruct additional wavelength bands using representation and sampling matrices, allowing for higher wavelength resolutions with fewer filters, eliminating the need for low-end and high-end filters, and leveraging correlation between adjacent bands for enhanced spectral resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple interpolation methods are used to reconstruct wavelength bands, then the system is easy to operate, but the reconstruction accuracy is insufficient
Solution Approach 1:
The patent transforms the reconstruction problem from direct spatial interpolation to a transformed domain problem using discrete cosine transform (DCT). By changing the parameter space from spatial domain to frequency domain, the method enables accurate reconstruction of wavelength bands while maintaining computational efficiency. The DCT coefficients are sparsely populated and then inverse transformed to obtain reconstructed bands, achieving both accuracy and operational simplicity.
Solution Approach 2:
The patent replaces traditional mechanical interpolation algorithms with a mathematical transformation approach using DCT. Instead of using complex mechanical or iterative optimization algorithms, the system substitutes a deterministic mathematical transform that provides both accuracy and computational efficiency, resolving the contradiction between operational simplicity and reconstruction accuracy.
2Manufacturing precision
If a large number of filters are used to capture more wavelength bands, then the spectral resolution is improved, but the hardware cost and device complexity increase
Solution Approach 1:
The patent creates virtual copies of wavelength bands through mathematical reconstruction rather than physical filters. By using DCT-based reconstruction, the system generates additional wavelength bands that did not physically exist in the captured data, effectively copying spectral information from measured bands to unmeasured bands. This eliminates the need for additional physical filters while maintaining spectral resolution.
Solution Approach 2:
The patent transitions from a one-dimensional filter selection problem to a multi-dimensional transformation problem. Instead of adding filters along the wavelength dimension, the system introduces a transformation domain (DCT coefficients) as an additional dimension, allowing spectral information to be reconstructed through mathematical operations in this new dimension rather than through additional physical components.
3Device complexity
If fewer filters are used to reduce hardware cost, then the device complexity is reduced, but the number of wavelength bands is insufficient
Solution Approach 1:
The patent performs preliminary action by capturing a limited set of wavelength bands with fewer filters, then uses DCT transformation to pre-compute reconstruction coefficients that enable generation of additional bands. This preliminary capture followed by mathematical expansion allows the system to obtain full spectral coverage without requiring all filters to be present during operation, reducing hardware complexity while maintaining the quantity of wavelength bands.
4Manufacturing precision
If fixed factory filters are used, then the manufacturing precision is ensured, but the adaptability to different wavelength requirements is limited
Solution Approach 1:
The patent introduces dynamics by replacing fixed factory filters with a dynamic reconstruction system. Instead of static physical filters that cannot be changed, the system uses computationally generated filters through DCT-based reconstruction. This allows the effective filter wavelengths to be dynamically adjusted by changing reconstruction parameters or basis functions, providing adaptability while maintaining manufacturing precision through controlled mathematical transformations.
Data Source
AI summary
What is disclosed is a system and method for image reconstruction using a compressed sensing framework to increase the number of wavelength bands in hyperspectral video systems. The present method utilizes a restricted representation matrix and sampling matrix to reconstruct bands to a very large number without losing information content. Reference multi-band image vectors are created and those vectors are processed in a block-wise form to obtain custom orthonormal representation matrices. A sampling matrix is also constructed offline in the factory. The compressed sensing protocol is applied using a l1-norm optimization (or relaxation) algorithm to reconstruct large number of wavelength bands with each band being interspersed within the band of interest that are not imaged. The teaching hereof leads to very large number of bands without increasing the hardware cost.