Method and apparatus for intra prediction with multiple weighted references
A method of performing intra prediction for encoding or decoding uses multiple layers of reference samples. The layers are formed into reference arrays that are used by a function, such as a weighted combination, to form a final prediction. The prediction is used in encoding or decoding a block of video data. The weights can be determined in a number of ways, and for a given prediction mode, the same weights, or different weights can be used for all pixels in a target block. If the weights are varied, they can depend on the distance of the target pixel from reference arrays.
1 . A method, comprising:
obtaining a plurality of reference arrays based on a plurality of reconstructed samples neighboring a block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the block of video data;
determining a prediction direction;
obtaining a first predicted sample from a first reference array of the plurality of reference arrays based on the prediction direction;
obtaining a second predicted sample from an interpolation of a first reference sample and a second reference sample from a second reference array of the plurality of reference arrays based on the prediction direction;
obtaining an intra prediction value for a target pixel of the block of video using a weighted combination of at least the first predicted sample and the second predicted sample, wherein the first reference array is closer to the block than the second reference array, and wherein a first weight applied to the first predicted sample is greater than a second weight applied to the second predicted sample; and
encoding the block of video data using the intra prediction value for the target pixel.
2 . The method of claim 1 , wherein each of the plurality of reference arrays is included in a respective layer comprising a row of pixels above the block of video data and a column of pixels to the left of the block of video data.
3 . The method of claim 1 , further comprising:
obtaining the first predicted sample from an interpolation of a third reference sample and a fourth reference sample from the first reference array.
4 . The method of claim 1 , wherein the first reference sample and the second reference sample are adjacent in the second reference array.
5 . The method of claim 1 , wherein the block is a luma block.
6 . A method, comprising:
obtaining a plurality of reference arrays based on a plurality of decoded samples neighboring a block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the block of video data;
determining a prediction direction;
obtaining a first predicted sample from an interpolation of a first reference sample and a second reference sample from a first reference array of the plurality of reference arrays based on the prediction direction;
obtaining a second predicted sample from a second reference array of the plurality of reference arrays based on the prediction direction;
obtaining an intra prediction value for a target pixel of the block of video using a weighted combination of at least the first predicted sample and the second predicted sample, wherein the first reference array is closer to the block than the second reference array, and wherein a first weight applied to the first predicted sample is greater than a second weight applied to the second predicted sample; and
decoding the block of video using the intra prediction value for the target pixel.
7 . The method of claim 6 , wherein each of the plurality of reference arrays is included in a respective layer comprising a row of pixels above the block of video data and a column of pixels to the left of the block of video data.
8 . The method of claim 6 , further comprising:
obtaining the second predicted sample from an interpolation of a third reference sample and a fourth reference sample from the second reference array.
9 . The method of claim 6 , wherein the first reference sample and the second reference sample are adjacent in the first reference array.
10 . The method of claim 6 , wherein the block is a luma block.
11 . An apparatus, comprising:
a processor and a memory which are configured to:
obtain a plurality of reference arrays based on a plurality of reconstructed samples neighboring a block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the block of video data;
determine a prediction direction;
obtain a first predicted sample from a first reference array of the plurality of reference arrays based on the prediction direction;
obtain a second predicted sample from an interpolation of a first reference sample and a second reference sample from a second reference array of the plurality of reference arrays based on the prediction direction;
obtain an intra prediction value for a target pixel of the block of video using a weighted combination of at least the first predicted sample and the second predicted sample, wherein the first reference array is closer to the block than the second reference array, and wherein a first weight applied to the first predicted sample is greater than a second weight applied to the second predicted sample; and
encode the block of video data using the intra prediction value for the target pixel.
12 . The apparatus of claim 11 , wherein each of the plurality of reference arrays is included in a respective layer comprising a row of pixels above the block of video data and a column of pixels to the left of the block of video data.
13 . The apparatus of claim 11 , wherein the processor and the memory are configured to obtain the first predicted sample from an interpolation of a third reference sample and a fourth reference sample from the first reference array.
14 . The apparatus of claim 11 , wherein the first reference sample and the second reference sample are adjacent in the second reference array.
15 . The apparatus of claim 11 , wherein the block is a luma block.
16 . An apparatus, comprising:
a processor and a memory which are configured to:
obtain a plurality of reference arrays based on a plurality of decoded samples neighboring a block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the block of video data;
determine a prediction direction;
obtain a first predicted sample from an interpolation of a first reference sample and a second reference sample from a first reference array of the plurality of reference arrays based on the prediction direction;
obtain a second predicted sample from a second reference array of the plurality of reference arrays based on the prediction direction;
obtain an intra prediction value for a target pixel of the block of video using a weighted combination of at least the first predicted sample and the second predicted sample, wherein the first reference array is closer to the block than the second reference array, and wherein a first weight applied to the first predicted sample is greater than a second weight applied to the second predicted sample; and
decode the block of video data using the intra prediction value for the target pixel of the block of video data.
17 . The apparatus of claim 16 , wherein each of the plurality of reference arrays is included in a respective layer comprising a row of pixels above the block of video data and a column of pixels to the left of the block of video data.
18 . The apparatus of claim 16 , wherein the processor and the memory are configured to obtain the second predicted sample from an interpolation of a third reference sample and a fourth reference sample from the second reference array.
19 . The apparatus of claim 16 , wherein the first reference sample and the second reference sample are adjacent in the first reference array.
20 . The apparatus of claim 16 , wherein the block is a luma block.
21 . A method, comprising:
obtaining a plurality of reference arrays based on a plurality of reconstructed samples neighboring a chroma block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the chroma block of video data;
obtaining an intra prediction value for a target pixel of the block of video using a weighted combination of a set of predicted samples from a set of the plurality of reference arrays, wherein a set of weights used for the weighted combination decrease as respective distances between the target pixel and the respective reference arrays of the set of reference arrays increase; and
encoding the chroma block of video data using the intra prediction value for the target pixel.
22 . The method of claim 21 , wherein the set of the plurality of reference arrays includes one chroma reference array for every two luma reference layers.
23 . The method of claim 21 , wherein a respective predicted sample is interpolated from two adjacent reference samples of a respective reference array based on a prediction direction for intra prediction.
24 . The method of claim 21 , wherein the set of the plurality of reference arrays is dependent on a prediction direction for intra prediction.
25 . A method, comprising:
obtaining a plurality of reference arrays based on a plurality of reconstructed samples neighboring a chroma block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the chroma block of video data;
obtaining an intra prediction value for a target pixel of the chroma block of video using a weighted combination of a set of predicted samples from a set of the plurality of reference arrays, wherein a set of weights used for the weighted combination decrease as respective distances between the target pixel and the respective reference arrays of the set of reference arrays increase; and
decoding the chroma block of video data using the intra prediction value for the target pixel.
26 . The method of claim 25 , wherein the set of the plurality of reference arrays includes one chroma reference array for every two luma reference layers.
27 . The method of claim 25 , wherein a respective predicted sample is interpolated from two adjacent reference samples of a respective reference array based on a prediction direction for intra prediction.
28 . The method of claim 25 , wherein the set of the plurality of reference arrays is dependent on a prediction direction for intra prediction.
29 . A apparatus, comprising:
a processor and a memory which are configured to:
obtain a plurality of reference arrays based on a plurality of reconstructed samples neighboring a chroma block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the chroma block of video data;
obtain an intra prediction value for a target pixel of the chroma block of video using a weighted combination of a set of predicted samples from a set of the plurality of reference arrays, wherein a set of weights used for the weighted combination decrease as respective distances between the target pixel and the respective reference arrays of the set of reference arrays increase; and
encode the chroma block of video data using the intra prediction value for the target pixel.
30 . The apparatus of claim 29 , wherein the set of the plurality of reference arrays includes one chroma reference array and two luma reference layers.
31 . The apparatus of claim 29 , wherein a respective predicted sample is interpolated from two adjacent reference samples of a respective reference array based on a prediction direction for intra prediction.
32 . The apparatus of claim 29 , wherein the set of the plurality of reference arrays is dependent on a prediction direction for intra prediction.
33 . A apparatus, comprising:
a processor and a memory which are configured to:
obtain a plurality of reference arrays based on a plurality of reconstructed samples neighboring a chroma block of video data, wherein each respective reference array of the plurality of reference arrays is adjacent to at least one other reference array of the plurality of reference arrays, and one of the plurality of reference arrays borders the chroma block of video data;
obtain an intra prediction value for a target pixel of the chroma block of video using a weighted combination of a set of predicted samples from a set of the plurality of reference arrays, wherein a set of weights used for the weighted combination decrease as respective distances between the target pixel and the respective reference arrays of the set of reference arrays increase; and
decode the chroma block of video data using the intra prediction value for the target pixel.
34 . The apparatus of claim 33 , wherein set of the plurality of reference arrays includes one chroma reference array and two luma reference layers.
35 . The apparatus of claim 33 , wherein a respective predicted sample is interpolated from two adjacent reference samples of a respective reference array based on a prediction direction for intra prediction.
36 . The apparatus of claim 33 , wherein the set of the plurality of reference arrays is dependent on a prediction direction for intra prediction.