Adaptive filtering of visual image using auxiliary image information
AssigneeYesVideo, Inc.
InventorMichele M. Covell
Life of this patent
Estimating the term…
Abstract
An adaptive filter is used to filter a target version of a visual image that is produced by processing an original version of the visual image, the characteristics of the adaptive filter being determined in accordance with one or more characteristics of the original version of the visual image. The orientation and/or strength of filtering of the adaptive filter can be adjusted based on local properties of the original image, which can enable the adaptive filter to avoid introducing blurring across true edges in an image.
Claims35
A method for adaptively filtering a target version of a visual image that has been produce by processing an original version of the visual image, comprising the steps of:
- filtering the target version of the visual image with each of a plurality of directional filters;
- weighting the output of each of the plurality of directional filters, wherein the step of weighting further comprises the steps of:
- filtering the original version of the visual image with each of the plurality of directional filters;
- comparing, for each directional filter, the filtered original version of the visual image produced by that directional filter to the original version of the visual image;
- and
- establishing the weight of the output from each directional filter in accordance with the difference between a prediction error for that directional filter and a minimum prediction error, wherein the prediction error is the difference between the filtered original version of the visual image produced by that directional filter and the original version of the visual image;
- and
- combining the weighted outputs from all of the directional filters to produce a filtered version of the visual image.
A method for adaptively filtering a target version of a visual image that has been produced by processing an original version of the visual image, comprising the steps of:
- identifying a characteristic of the original version of the visual image;
- filtering the target version of the visual image with each of a plurality of directional filters;
- weighting the output of each of the plurality of directional filters in accordance with the characteristic of the original version of the visual image;
- averaging the outputs from all of the plurality of directional filters;
- and
- combining the averaged output with the weighted outputs from all of the directional filters, in accordance with the characteristic of the original version of the visual image, to produce a filtered version of the visual image, wherein the step of combining the with all further comprises the steps of:
- filtering the original version of the visual image with each of the plurality of directional filters;
- comparing the filtered original version of the visual image to the original version of the visual image;
- and
- combining the averaged output with the weighted outputs in accordance with a prediction error between the filtered original version of the visual image and the original version of the visual image for that directional filter.
Apparatus for adaptively filtering a target version of a visual image that has been produced by processing an original version of the visual image, comprising:
- means for identifying a characteristic of the original version of the visual image;
- and
- means for filtering the target version of the visual image with each of a plurality of directional filters;
- means for weighting the output of each of a plurality of directional filters in accordance with the characteristic of the original version of the visual image;
- means for averaging the outputs from all of the plurality of directional filters;
- and
- means for combining the averaged output with the weighted outputs from all of the directional filters, in accordance with the characteristic of the original version of the visual image, to produce a filtered version of the visual image wherein the means for combining further comprises:
- means for filtering the original version of the visual image will each of the plurality of directional filters;
- means for comparing the filtered original version of the visual image to the original version of the visual image;
- and
- means for combining the averaged output with the weighted outputs in accordance with a prediction error between the filtered original version of the visual image and the original version of the visual image for that directional filter.
A computer readable medium or media encoded with one or more computer programs for adaptively filtering a target version of a visual image that has been produced by processing an original version of the visual image, comprising:
- instructions for identifying a characteristic of the original version of the visual image;
- instructions for filtering the target version of the visual image with each of a plurality of directional filters;
- instructions for weighting the output of each of a plurality of directional filters in accordance with the characteristic of the original version of the visual image;
- instructions for averaging the outputs from all of the plurality of directional filters;
- in
- instructions for combining the averaged output with the weighted outputs from all of the directional filters, in accordance with the characteristic of the original version of the visual image, to produce a filtered version of the visual image, wherein the instructions for combining further comprise:
- instructions for filtering the original version of the visual image with each of the plurality of directional filters;
- instructions for comparing the filtered original version of the visual image to the original version of the visual image;
- and
- instructions for combining the averaged outputs with the weighted outputs in accordance with a prediction error between the filtered original version of the visual image and the original version of the visual image for that directional filter.
A computer readable medium or media encoded with one or more computer programs for adaptively filtering a target version of a visual image that has been produced by processing an original version of the visual image, comprising:
- instructions for filtering the target version of the visual image with each of a plurality of directional filters;
- instructions for weighting the output of each of the plurality of directional filters, wherein the instructions for weighting further comprise:
- instructions for filtering the original version of the visual image with each of the plurality of directional filters;
- instructions for comparing, for each directional filter, the filtered original version of the visual image produced by that directional filter to the original version of the visual image;
- and
- instructions for establishing the weight of the output from each directional filter in accordance with the difference between a prediction error for that directional filter and a minimum prediction error, wherein the prediction error is the difference between the filtered original version of the visual image produced by that directional filter and the original version of the visual image;
- and
- instructions for combining the weighted outputs from all of the directional filters to produce a filtered version of the visual image.
Apparatus for adaptively filtering a target version of a visual image that has been produced by processing an original version of the visual image, comprising:
- means for filtering the target version of the visual image with each of a plurality of directional filters;
- means for weighting the output of each of the plurality of directional filters, wherein the means for weighting further comprises:
- means for filtering the original version of the visual image with each of the plurality of directional filters;
- means for comparing, for each direct directional filter, the filtered original version of the visual image produced by that directional filter to the original version of the visual image;
- and
- means for establishing the weight of the output from each directional filter in accordance with the difference between a prediction error for that directional filter and a minimum prediction error, wherein the prediction error is the difference between the filtered original version of the visual image produced by that directional filter and the original version of the visual image;
- and
- means for combining the weighted outputs from all of the directional filters to produce a filtered version of the visual image.
Dependencies are inferred from the claim text.