Back to results
US 9,179,148 B2Granted
Simplified bilateral intra smoothing filter
AssigneesFuturewei Technologies, Inc., Santa Clara University
InventorsGuichun Li, Lingzhi Liu, Changcai Lai, Nam Ling, Jianhua Zheng, Chen-Xiong Zhang
Life of this patent
Estimating the term…
Abstract
A method comprising receiving a plurality of reference pixels, computing a plurality of filter coefficients based on differences between a reference pixel and neighboring reference pixels in the plurality of reference pixels, and combining the filter coefficients with the reference pixel and the neighboring reference pixels to generate a filtered value, wherein the filtered value is used for intra prediction.
Claims11
A method comprising:
- receiving a plurality of reference pixels;
- computing a plurality of filter coefficients based on differences between a reference pixel and neighboring reference pixels in the plurality of reference pixels and based on distances between the reference pixel and the neighboring reference pixels;
- and
- combining the filter coefficients with the reference pixel and the neighboring reference pixels to generate a filtered value, wherein the filtered value is a linear combination of the reference pixel and the neighboring reference pixels and used for intra prediction,
- wherein the plurality of filter coefficients comprise three filter coefficients, wherein a first distance between the reference pixel and a left pixel is equal to one, wherein a second distance between the reference pixel and a right pixel is equal to one, and wherein at least one filter coefficient is based on a maximum value for the plurality of reference pixels, a degree of filtering, a first difference between the reference pixel value and the left pixel value, a second difference between the reference pixel value and the right pixel value, or any combination thereof, and
- wherein the filter coefficients are:
- [
- Max
- -
- k
- *
- DL
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- 2
-
- Max
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- Max
- -
- k
- *
- DR
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
- ]
- ,
- where Max represents the maximum value for the plurality of reference pixels up to 255, k is the degree of filtering between 1 and 6, DL is the first difference, and DR is the second difference.
55 more elements
An apparatus comprising:
- a processor configured to:
- receive a plurality of reference pixels;
- compute a plurality of filter coefficients based on differences between a reference pixel and neighboring reference pixels in the plurality of reference pixels and based on distances between the reference pixel and the plurality of neighboring reference pixels;
- and
- combine the filter coefficients with the reference pixel and the neighboring reference pixels to generate a filtered value, wherein the filtered value is a linear combination of the reference pixel and the neighboring reference pixels and used for intra prediction,
- wherein the plurality of filter coefficients comprise three filter coefficients, wherein a first distance between the reference pixel and a left pixel is equal to one, wherein a second distance between the reference pixel and a right pixel is equal to one, and wherein at least one filter coefficient is based on a maximum value for the plurality of reference pixels, a degree of filtering, a first difference between the reference pixel value and the left pixel value, a second difference between the reference pixel value and the right pixel value, or any combination thereof, and
- wherein the filter coefficients are:
- [
- Max
- -
- k
- *
- DL
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- 2
-
- Max
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- Max
- -
- k
- *
- DR
- 4
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
- ]
- ,
- where Max represents the maximum value for the plurality of reference pixels up to 255, k is the degree of filtering between 1 and 6, DL is the first difference, and DR is the second difference.
56 more elements
An apparatus comprising:
- at least one processor configured to:
- receive a plurality of reference pixels, wherein the reference pixels comprise a current pixel and a plurality of neighboring pixels;
- compute a plurality of bilateral filter coefficients based on:
- a range filter comprising a plurality of range filter coefficients, wherein at least one range filter coefficient is based on a difference between the current pixel and at least one neighboring pixel, and
- a domain filter comprising a plurality of domain filter coefficients, wherein at least one domain filter coefficient is based on a distance between the current pixel and at least one neighboring pixel;
- and
- combine the bilateral filter coefficients with the current pixel and the neighboring pixels to generate a filtered value, wherein the filtered value is a linear combination of the current pixel and the plurality of neighboring pixels and used for intra prediction,
- wherein the range filter is a 3-tap filter comprising three range filter coefficients, wherein at least one range filter coefficient is based on a maximum value for the plurality of reference pixels, a degree of filtering, a first difference between the current pixel value and a left pixel value, a second difference between the current pixel value and a right pixel value, or any combination thereof, and
- wherein the range filter coefficients are:
- [
- Max
- -
- k
- *
- DL
- 3
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- Max
- 3
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
-
- Max
- -
- k
- *
- DR
- 3
-
- Max
- -
- k
-
- (
- DL
- +
- DR
- )
- ]
- ,
- where Max is the maximum value for the plurality of reference pixels up to 255, k is the degree of filtering between 1 and 6, DL is a positive integer representing the first difference, and DR is a positive integer representing the second difference.
56 more elements
Dependencies are inferred from the claim text.