Image analysis – Color image processing – Color correction
Reexamination Certificate
1996-12-19
2001-02-27
Boudreau, Leo (Department: 2621)
Image analysis
Color image processing
Color correction
C382S162000, C382S166000, C382S240000, C382S248000, C358S518000
Reexamination Certificate
active
06195456
ABSTRACT:
BACKGROUND OF THE INVENTION
There are a number of processes that may be applied to an image such as descreening, compression, segmentation, etc, which use multiresolution analysis. In this case, the image is divided into a number of frequency sub-bands, the process is applied to some of the sub-bands, and the sub-bands are recombined to form the processed image.
For color images, the problem is that running all of these processes for each color separation, along with other color conversions, results in a large amount of computational overhead. A method of reducing the amount of computation is needed.
An improved color image printing method is proposed for combining the aforementioned multiresolution representation of an image with the color conversions required for high quality printing. This is achieved by dividing the color conversion into two steps: a computationally complex 3-dimensional printer color correction step, applied to a lower resolution sub image; and a computationally simpler color space transformation step, comprising primarily 1-D functions, applied to the full resolution image. The computational cost is thereby reduced by applying the more difficult computation, color correction, to the sub image which has fewer image points.
SUMMARY OF THE INVENTION
The color conversions normally required to translate images from a device independent color representation to a printer specific representation can be divided into two steps. Color space transformations (device RGB to device CMYK) are relatively simple transformations, and can be done by mostly 1-D table lookups, and a small number of operations. In contrast, color correction from a device independent specification, R′G′B′, to a set of device values, RGB that must be used for the specific printer to produce the desired colors, is a relatively computationally intensive process involving 3-D table lookups and interpolation. Thus, while the simpler 1-D color space transformations can be applied to the final full-resolution image, it would be an advantage if the computations in the 3-D color correction step could be reduced. The proposed solution is to apply the 3-D color correction to the much smaller number of pixels that already exist in a lower resolution image that is produced in the multiresolution analysis process. It has been determined that applying the color correction at a lower resolution results in a small loss of image quality that is not readily discernible in the final image while the amount of computation is significantly decreased.
One particular method of reducing the resolution of the image that is particularly suited to this use is wavelet transformation. In this process the image is passed through a number of filter stages, each of which separates the input into four bands. Each stage sends the lowest frequency band to the input of the next stage and outputs the other three bands as intermediate images. Color correction is applied to one or more of the lowest frequency bands, which have the lowest resolutions. Finally, all of the bands are recombined to a higher resolution image which now contains the color correction. The system is optimized by determining the lowest resolution to which the color correction can be applied with acceptable results.
REFERENCES:
patent: 5666475 (1997-09-01), Salesin et al.
Balasubramanian Thyagarajan
de Queiroz Ricardo L.
Boudreau Leo
Cunha Robert
Novik Dmitry A.
Xerox Corporation
LandOfFree
Multi resolution color correction using wavelet transforms does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Multi resolution color correction using wavelet transforms, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multi resolution color correction using wavelet transforms will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2572185