Multi-spectral image compression with bounded loss
AssigneeInternational Business Machines Corporation
InventorsBoon-Lock Yeo, Joan LaVerne Mitchell
Life of this patent
Estimating the term…
Abstract
A system and method for generating bounded-loss color transformation employed in the compression and decompression of multi-spectral images is provided. The bounded-loss color space transformations provide transformations from a first color space to a second color space and back to the first color space with bounded loss or error. The bounded loss or error allows for the near lossless compression and reconstruction of multi-spectral images transformed from a first color space to a second color space and ultimately back to the first color space.
Claims20
A computer-implemented method for compressing a digital multi-spectral image represented by a first color space, the method comprising the steps of:
- a) dividing a first set of first color space components into first and second paths;
- b-1) compressing the first color space components on the first path;
- b-2) dividing the compressed color space components into two paths;
- b-3) inputting one path of the compressed first color space components into an output function;
- b-4) reconstructing the other path of compressed first color space components;
- and
- b-5) transforming the reconstructed first color space components into a first set of second color space components using a first bounded-loss color space transformation;
- c) transforming the second set of first color space components on the second path into a second set of second color space components using a second bounded-loss color space transformation;
- d) inputting the differences between each of the first set of second color space components and the second set of second color space components into a difference function whereby the differences are bound by a predetermined range of values;
- e) compressing the output of the differences;
- and
- f) inputting the compressed differences into the output function.
A computer-implemented method for compressing a digital multi-spectral image represented by a first color space, the method comprising the steps of:
- a) dividing a first set of first color space components into first and second paths;
- b-1) transforming the first set of first color space components on the first path into a plurality of second color space components using a first bounded-loss color space transformation;
- b-2) compressing the second color space components and dividing the compressed second color space components into two paths;
- b-3) inputting one path of the compressed second color space components into an output function;
- b-4) reconstructing the other path of the compressed second color space components;
- b-5) transforming the reconstructed second color space components into a first set of third color space components using a second bounded-loss color space transformation;
- c) transforming the first color space components on the second path into a second set of third color space components;
- d) inputting the differences between each of the first set of third color space components and the second set of third color space components into a difference function whereby the differences are bound by a predetermined range of values;
- e) compressing the output of the differences of the first set of third color space components, and the second set of third color space components;
- and
- f) inputting the compressed differences into the output function.
A method for compressing digital multi-spectral images comprising the steps of:
- a) compressing a first set of first color space components to thereby generate first color space compressed data;
- b) reconstructing the first color space compressed data to a plurality of reconstructed first color space components;
- c) transforming the reconstructed first color space components to a first set of second color space components;
- d) determining component differences based on the differences between the second set of color space components and a second set of second color space components that have been transformed from a second set of first color space components without first being compressed and reconstructed;
- e) compressing the component differences of the second color space components;
- and
- f) incorporating the second color space compressed data and a compressed set of first color space data into a compressed data stream.
A method for compressing digital multi-spectral images comprising the steps of:
- a) transforming a first set of first color space components to a first set of second color space components
- a) compressing a first set of second color space components to thereby generate second color space compressed data;
- b) reconstructing the first set of second color space compressed data to a plurality of reconstructed second color space components;
- c) transforming the reconstructed first set of second color space components to a first set of third color space components;
- d) determining component differences based on the differences between the first set of third color space components and a second set of third color space components that have been transformed from a second set of first color space components without first being compressed and reconstructed;
- e) compressing the component differences of the third color space components;
- and
- f) incorporating the compressed differences of the third color space components and a compressed set of second color space data into a compressed data stream.
Dependencies are inferred from the claim text.