Method and system for compression of hyperspectral or multispectral imagery with a global optimal compression algorithm (GOCA)
Inventors
Assignees
NAVY GOVERNMENT OF United States, Secretary of • US Department of Navy
Publication Number
US-8432974-B2
Publication Date
2013-04-30
Expiration Date
2030-07-12
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A computer based method and system for compressing digital hyperspectral or multispectral image data. The method includes initially reducing the plurality of spectral bands of the hyperspectral data to a smaller number of spectral bands using principal component analysis, determining an optimum compression ratio for each of the smaller number of spectral bands for use in a wavelet transform, and subsequently compressing the smaller number of spectral bands spatially using the wavelet transform with the optimum compression ratios.
Core Innovation
The invention provides a computer-based method and system for compressing digital hyperspectral or multispectral image data by initially reducing the plurality of spectral bands of the hyperspectral data to a smaller number of spectral bands using principal component analysis (PCA). Then, an optimum compression ratio is determined for each of these smaller spectral bands for use in a wavelet transform, followed by spatial compression of these bands using the wavelet transform with the optimum compression ratios.
The method aims to optimize compression by minimizing the mean square error (MSE) in both the spectral and spatial dimensions. This involves minimizing the truncation error after the PCA transform and minimizing the spatial mean square error through a semi-empirical rate distortion function for non-Gaussian images, with compression ratios tailored individually for each PCA-derived spectral band to preserve spectral integrity and image quality.
The problem being addressed is the need for efficient compression of hyperspectral and multispectral image data, which typically involves large data volumes due to numerous spectral bands. Maintaining spectral and spatial image quality while achieving significant data compression ratios is challenging, especially for real-time transmission and storage. Existing methods either use single-stage compression without separating spectral and spatial redundancies or two-stage methods without optimized variable compression ratios. This invention improves upon these by providing a global optimal compression algorithm (GOCA) that separately optimizes compression in both spectral and spatial dimensions using PCA and wavelet transforms, thus balancing the trade-offs between compression ratio, processing requirements, and image quality preservation.
Claims Coverage
The patent includes three independent claims covering a method, a computer-readable medium, and a system for hyperspectral data compression with global optimum compression ratios.
Compression method using PCA and wavelet transform with optimum compression ratios
A computer-based method that compresses digital hyperspectral data by reducing spectral bands using principal component analysis (PCA), determining an optimum compression ratio for each reduced spectral band for use in a wavelet transform, and spatially compressing those bands with the optimum ratios, wherein the optimum compression ratios minimize mean square error.
Programmed instructions on a non-transitory medium implementing optimum compression
A computer-readable non-transitory medium storing instructions to compress received digital hyperspectral image data by reducing spectral bands via PCA, determining optimum compression ratios for each smaller spectral band for use in a wavelet transform, and compressing spatially using the wavelet transform with those optimum ratios, including minimizing mean square error.
System comprising a receiver and computer system executing optimized compression
A system for compressing digital hyperspectral data comprising a receiver for receiving said data and a computer system programmed to reduce spectral bands using PCA, determine optimum compression ratios for each smaller spectral band for use in a wavelet transform, and spatially compress the bands with these optimum ratios, minimizing overall mean square error.
The independent claims collectively cover a comprehensive approach to hyperspectral image data compression that uses PCA for spectral reduction, optimization of compression ratios per spectral component based on minimizing error, and subsequent spatial compression via wavelet transform, implemented as a method, program storage medium, and system.
Stated Advantages
Provides very high compression ratios with low distortion by optimizing compression ratios individually for each spectral component after PCA.
Maintains spectral integrity and spatial image quality by separately optimizing compression in spectral and spatial dimensions.
Improves compression efficiency significantly compared to uniform compression methods and existing standards such as JPEG 2000, achieving up to 250% better compression ratio for the same signal-to-noise ratio in test scenarios.
Enables practical real-time transmission and storage of large hyperspectral or multispectral data sets with preserved image quality.
Documented Applications
Compression of hyperspectral or multispectral data acquired from satellite and aerial imagery systems.
Ground-based compression of multispectral or hyperspectral images including three band red-green-blue imagery and medical images.
Use in remote sensing applications involving geological, oceanographic, agricultural, ecological, medical imagery, and atmospheric science data.
Interested in licensing this patent?