Filtering with side-information using contextually-designed filters
AssigneeGoogle Inc.
InventorsOnur Guleryuz, Debargha Mukherjee
Life of this patent
Estimating the term…
Abstract
Obtaining a restored frame from a degraded frame includes obtaining, for a pixel of the degraded frame, magnitude features based on a first window centered at the pixel. A cardinality N of the magnitude features is at least 1. The magnitude features are used to obtain a pixel-adaptive filter. The pixel-adaptive filter is applied to the pixel to obtain a pixel of the restored frame.
Claims20
A method for obtaining a restored frame from a degraded frame, comprising:
- obtaining, for a pixel of the degraded frame, magnitude features based on a first window centered at the pixel, wherein a cardinality N of the magnitude features is at least 1;
- using the magnitude features to obtain a pixel-adaptive filter, wherein using the magnitude features to obtain the pixel-adaptive pixel comprises:
- obtaining, for the pixel, N classification features, wherein each of the N classification features corresponds to an average of respective magnitude features of pixels of a second window that is centered at the pixel;
- and
- applying the pixel-adaptive filter to the pixel to obtain a pixel of the restored frame.
A device, comprising:
- a processor, the processor configured to:
- obtain, for a pixel of a degraded frame, magnitude features based on a first window centered at the pixel, wherein a cardinality N of the magnitude features is at least 1;
- use the magnitude features to obtain a pixel-adaptive filter, wherein to use the magnitude features to obtain the pixel-adaptive pixel comprises to:
- obtain, for the pixel, N classification features, wherein each of the N classification features corresponds to an average of respective magnitude features of pixels of a second window that is centered at the pixel;
- and
- apply the pixel-adaptive filter to the pixel to obtain a pixel of a restored frame.
A non-transitory computer-readable storage medium, comprising executable instructions that, when executed by a processor, cause performance of operations for obtaining a restored frame from a degraded frame, comprising:
- obtaining, for a pixel of the degraded flame, magnitude features based on a first window centered at the pixel, wherein a cardinality N of the magnitude features is at least 1:
- using the magnitude features to obtain a pixel-adaptive filter, wherein using the magnitude features to obtain the pixel-adaptive pixel comprises:
- obtaining, for the pixel, N classification features, wherein each of the N classification features corresponds to an average of respective magnitude features of pixels of a second window that is centered at the pixel;
- and
- applying the pixel-adaptive filter to the pixel to obtain a pixel of the restored frame.
Dependencies are inferred from the claim text.