Method to remove the spectral components of illumination and background from multi-spectral and hyper-spectral images
AssigneeCenter for Quantitative Cytometry
InventorsAbraham Schwartz, Phillip Sherman, Peter Ingram
Life of this patent
Estimating the term…
Abstract
The present invention is a method of removing the illumination and background spectral components thus isolating spectra from multi-spectral and hyper-spectral data cubes. The invention accomplishes this by first balancing a reference and sample data cubes for each spectra associated with each location, or pixel/voxel, in the spatial image. The set of residual spectra produced in the balancing step is used to obtain and correct a new set of reference spectra that is used to remove the illumination and background components in a sample data cube.
Claims19
A method for removing spectral components of illumination energy and background from multi-spectral and hyper-spectral data cubes containing spatial and spectral images, the method comprising:
- aligning a data cube of a sample spectral camera with a data cube of a reference spectral camera;
- obtaining an empty sample data cube (xse, yse, λse) from said sample spectral camera directed at a first area and an empty reference data cube (xre, yre, λre) from said reference spectral camera directed at a second area;
- subtracting a spectral intensity of each pixel of said empty reference data cube (xre, yre, λre) from a spectral intensity of a corresponding pixel of said empty sample data cube (xse, yse, λse) to obtain a residual spectral data cube (xR, yR, λR);
- adding the spectral intensity of each pixel of the residual spectral data cube (xR, yR, λR) to the spectral intensity of a corresponding pixel of said empty reference data cube (xre, yre, λre) to obtain a resulting reference data cube [(xre, yre, λre)+(xR, yR, λR)];
- subtracting the intensity of each pixel of the resulting reference data cube [(xre, yre, λre)+(xR, yR, λR)] from the intensity of a corresponding pixel of said empty sample data cube (xse, yse, λse) to obtain a zero order spectra data cube where all spectral intensities throughout the data cube are zero;
- obtaining a sample data cube (xs, ys, λs) from said sample spectral camera directed at a third area and a reference data cube (xr, yr, λr) from said reference spectral camera directed at a fourth area;
- adding the spectral intensity of each pixel of the residual spectral data cube (xR, yR, λR) to the spectral intensity of a corresponding pixel of said reference data cube (xr, yr, λr) to obtain a new resulting reference data cube [(xr, yr, λr)+(xR, yR, λR)];
- and
- subtracting the spectral intensity of each pixel of the resulting reference data cube [(xr, yr, λr)+(xR, yR, λR)] from the intensity of a corresponding pixel of said sample data cube (xs, ys, λs) to obtain a noise-free data cube, effectively removing from said noise-free data cube at least one of:
- spectral illumination components, aerosol components, and background components.
A method for removing spectral components of illumination energy and background from multi-spectral and hyper-spectral data cubes containing spatial and spectral images, the method comprising:
- aligning a data cube generated from a sample spectral source with a data cube generated from a reference spectral source;
- generating an empty sample data cube (xse, yse, λse) from said sample spectral source directed at a first area and an empty reference data cube (xre, yre, λre) from said reference spectral source directed at a second area;
- subtracting a spectral intensity of each pixel of said empty reference data cube (xre, yre, λre) from a spectral intensity of a corresponding pixel of said empty sample data cube (xse, yse, λse) to obtain a residual spectral data cube (xR, yR, λR);
- adding the spectral intensity of each pixel of the residual spectral data cube (xR, yR, λR) to the spectral intensity of a corresponding pixel of said empty reference data cube (xre, yre, λre) to obtain a resulting reference data cube [(xre, yre, λre)+(xR, yR, λR)];
- subtracting the intensity of each pixel of the resulting reference data cube [(xre, yre, λre)+(xR, yR, λR)] from the intensity of a corresponding pixel of said empty sample data cube (xse, yse, λse) to obtain a zero order spectra data cube where all spectral intensities throughout the data cube are zero;
- generating a sample data cube (xs, ys, λs) from said sample spectral source directed at a third area and a reference data cube (xr, yr, λr) from said reference spectral source directed at a fourth area;
- adding the spectral intensity of each pixel of the residual spectral data cube (xR, yR, λR) to the spectral intensity of a corresponding pixel of said reference data cube (xr, yr, λr) to obtain a new resulting reference data cube [(xr, yr, λr)+(xR, yR, λR)];
- and
- subtracting the spectral intensity of each pixel of the new resulting reference data cube [(xr, yr, λr) +(xR, yR, λR)] from the intensity of a corresponding pixel of said sample data cube (xs, ys, λs) to obtain a noise-free data cube, effectively removing from said noise-free data cube at least one of:
- spectral illumination components, aerosol components, and background components.
Dependencies are inferred from the claim text.