Method of compressing JPEG files using a conditional transform
AssigneeXerox Corporation
InventorRicardo L. de Queiroz
Life of this patent
Estimating the term…
Abstract
A method for compressing digital image data to improve the efficiency of serial data transmission is disclosed. More specifically, the present invention accomplishes image compression by performing the most complex portions of a standard compression technique on a subset of the originally provided data utilizing a modified two-dimensional discrete cosine transform. The invention includes a fast JPEG compressor using a Haar transform with a conditional transform.
Claims25
A method of improving the speed and efficiency of electronic data compression, comprising:
- obtaining input image data which includes discrete values that represent light intensity in an image;
- applying a first transform to the input image data to produce a first transform result;
- comparing the first transform result to a threshold;
- applying a second transform to the first transform result to generate approximation data, if the first transform result has a pre-selected relationship to the threshold;
- and
- substituting a zero value for the first transform result to generate approximation data, if the first transform result does not have the pre-selected relationship to the threshold.
A method of transmitting a facsimile of an original image from a sending location to a receiving location, comprising:
- acquiring the original image and generating digital image data that includes pixel values which represent the light intensity of the original image;
- applying a first transform to the input image data to produce a first transform result;
- comparing the first transform result to a threshold;
- applying a second transform to the first transform result to generate approximation data, if the first transform result has a pre-selected relationship to the threshold;
- substituting a zero value for the first transform result to generate approximation data, if the first transform result does not have the pre-selected relationship to the threshold;
- and
- deriving output image data from the approximation data.
Dependencies are inferred from the claim text.