Intra interpolation filter for multi-line intra prediction
A method and apparatus for selecting an intra interpolation filter for multi-line intra prediction based on a reference line index for decoding a video sequence includes identifying a set of reference lines associated with a coding unit. A first type of interpolation filter is applied to reference samples included in a first reference line, of the set of reference lines, that is adjacent to the coding unit to generate a first set of prediction samples based on the first reference line being associated with a first reference line index. A second type of interpolation filter is applied to reference samples included in a second reference line, of the set of reference lines, that is non-adjacent to the coding unit to generate a second set of prediction samples based on the second reference line being associated with a second reference line index.
1 . A method for decoding a video sequence, comprising:
identifying a set of reference lines associated with a coding unit;
applying a first filter selected from an edge smoothing type of filter or an edge preserving type of filter to reference samples included in a first reference line of the set of reference lines based on the first reference line being associated with a first reference line index, wherein the first reference line index indicates that the first reference line is adjacent to the coding unit, and wherein the edge smoothing type of filter comprises a linear interpolation filter having only positive or zero filter coefficients; and
applying a second filter selected from only the edge preserving type of filter to reference samples included in a second reference line of the set of reference lines based on the second reference line being associated with a second reference line index, wherein the second reference line index indicates that the second reference line is non-adjacent to the coding unit, wherein the edge preserving type of filter comprises a linear interpolation filter having at least one negative filter coefficient, a non-linear interpolation filter, a Cubic filter, a discrete cosine transform (DCT)-based filter, a polynomial-based interpolation filter, a bilateral filter, and a Hermite interpolation filter.
2 . The method of claim 1 , wherein the edge smoothing type of filter comprises a bilinear filter and a Gaussian filter.
3 . The method of claim 1 , wherein applying the first filter comprises applying a first type of edge preserving filter, and wherein applying the second filter comprises applying a second type of edge preserving filter that is different than the first type of edge preserving filter.
4 . The method of claim 1 , wherein applying the first filter comprises applying an M-tap edge preserving filter, wherein applying the second filter comprises applying an N-tap interpolation filter, and wherein M is not equal to N.
5 . The method of claim 1 , wherein applying the first filter comprises applying a first edge preserving filter associated with a first set of filter coefficients, and wherein applying the second filter comprises applying a second edge preserving filter associated with a second set of filter coefficients that is different than the first set of filter coefficients.
6 . The method of claim 1 , wherein applying the first filter comprises applying an edge preserving filter including fixed filter coefficients.
7 . The method of claim 1 , wherein applying the first filter comprises applying an edge smoothing filter in association with wide angle modes of the first reference line, and wherein applying the second filter comprises applying an edge preserving filter that is applied to wide angle modes of the second reference line.
8 . The method of claim 1 , further comprising:
generating a first set of prediction samples by applying the first filter; and
generating a second set of prediction samples by applying the second filter.
9 . A method for encoding a video sequence, comprising:
identifying a set of reference lines associated with a coding unit;
selecting a first filter from an edge smoothing type of filter or an edge preserving type of filter to be applied to reference samples included in a first reference line of the set of reference lines based on the first reference line being associated with a first reference line index, wherein the first reference line index indicates that the first reference line is adjacent to the coding unit, and wherein the edge smoothing type of filter comprises a linear interpolation filter having only positive or zero filter coefficients;
generating a first set of prediction samples based on the first reference line;
selecting a second filter from only the edge preserving type of filter, not the edge smoothing type of filter, to be applied to reference samples included in a second reference line of the set of reference lines based on the second reference line being associated with a second reference line index, wherein the second reference line index indicates that the second reference line is non-adjacent to the coding unit, wherein the edge preserving type of filter comprises a linear interpolation filter having at least one negative filter coefficient, a non-linear interpolation filter, a Cubic filter, a discrete cosine transform (DCT)-based filter, a polynomial-based interpolation filter, a bilateral filter, and a Hermite interpolation filter; and
generating a second set of prediction samples based on the second reference line, wherein only most probable modes (MPMs) are allowed for the second reference line.
10 . The method of claim 9 , wherein the edge smoothing type of filter comprises a bilinear filter and a Gaussian filter.
11 . The method of claim 9 , wherein:
selecting the first filter comprises selecting a first type of edge preserving filter; and
selecting the second filter comprises selecting a second type of edge preserving filter that is different than the first type of edge preserving filter.
12 . The method of claim 9 , wherein:
applying the first filter comprises applying an M-tap edge preserving filter;
applying the second filter comprises applying an N-tap interpolation filter; and
M is not equal to N.
13 . The method of claim 9 , wherein:
selecting the first filter comprises selecting a first edge preserving filter associated with a first set of filter coefficients; and
selecting the second filter comprises selecting a second edge preserving filter associated with a second set of filter coefficients that is different than the first set of filter coefficients.
14 . The method of claim 9 , wherein selecting the first filter comprises selecting an edge preserving filter including fixed filter coefficients.
15 . The method of claim 9 , wherein:
selecting the first filter comprises selecting an edge smoothing filter in association with wide angle modes of the first reference line; and
selecting the second filter comprises selecting an edge preserving filter that is applied to wide angle modes of the second reference line.
16 . A method of processing visual media data, the method comprising:
performing a conversion between a visual media file and a bitstream of a visual media data according to a format rule,
wherein the bitstream comprises a plurality of coding units; and
wherein the format rule specifies that:
a set of reference lines associated with a coding unit of the plurality of coding units is to be identified;
a first filter, selected from an edge smoothing type of filter or an edge preserving type of filter, is to be applied to reference samples included in a first reference line of the set of reference lines based on the first reference line being associated with a first reference line index, wherein the first reference line index indicates that the first reference line is adjacent to the coding unit, and wherein the edge smoothing type of filter comprises a linear interpolation filter having only positive or zero filter coefficients; and
a second filter, selected from only the edge preserving type of filter, is to be applied to reference samples included in a second reference line of the set of reference lines based on the second reference line being associated with a second reference line index, wherein the second reference line index indicates that the second reference line is non-adjacent to the coding unit, wherein the edge preserving type of filter comprises a linear interpolation filter having at least one negative filter coefficient, a non-linear interpolation filter, a Cubic filter, a discrete cosine transform (DCT)-based filter, a polynomial-based interpolation filter, a bilateral filter, and a Hermite interpolation filter.
17 . The method of claim 16 , wherein the edge smoothing type of filter comprises a bilinear filter and a Gaussian filter.
18 . The method of claim 16 , wherein:
applying the first filter comprises applying a first type of edge preserving filter, and
applying the second filter comprises applying a second type of edge preserving filter that is different than the first type of edge preserving filter.
19 . The method of claim 16 , wherein:
applying the first filter comprises applying an M-tap edge preserving filter,
applying the second filter comprises applying an N-tap interpolation filter, and
M is not equal to N.
20 . The method of claim 16 , wherein applying the first filter comprises applying an edge preserving filter including fixed filter coefficients.