Video bitstream coding
A method of decoding JVET video includes receiving a bitstream indicating how a coding tree unit was partitioned into coding units, and parsing said bitstream to generate at least one predictor based on an intra prediction mode signaled in the bitstream. The predictor may be generated by interpolating neighboring pixels for each pixel within a coding block. The computation may be more accurate by deriving a value for a bottom right neighboring pixel.
1. A method for using planar mode to predict pixel values for a current coding block having height H, width W, and a top-left (TL) pixel within the current coding block is defined by coordinates x=0 and y=0, where constructed or reconstructed neighboring pixels form a row of horizontal boundary pixels on an upper side of the current coding block and constructed or reconstructed neighboring pixels form a column of vertical boundary pixels on a left side of the current coding block, the neighboring pixels including a top right neighboring pixel, a top left neighboring pixel, and a bottom left neighboring pixel, the method configured to:
calculate an intensity value of a bottom right neighboring pixel of the current coding block;
use an intensity value of a neighboring pixel from the column of vertical boundary pixels on the left side of the current coding block and the calculated intensity value of the bottom right neighboring pixel, calculate intensity values for the neighboring pixels in a row along the lower side of the current coding block;
use an intensity value of a neighboring pixel from the row of horizontal boundary pixels on the upper side of the current coding block and the calculated intensity value of the bottom right neighboring pixel, calculate intensity values for the neighboring pixels in a column along the right side of the current coding block;
calculate a first predictor for a pixel at coordinate (x, y) in the current coding block by interpolating between an intensity value of a respective vertical neighboring pixel in the row of neighboring pixels along the upper side of the current coding block and at least one of the calculated intensity values for the neighboring pixels in the row along the lower side of the current coding block;
calculate a second predictor for the pixel at coordinate (x, y) in the current coding block by interpolating between an intensity value of a respective horizontal neighboring pixel from the row of neighboring pixels along the left side of the current coding block and at least one of the calculated intensity values for the neighboring pixels in the column along the right side of the current coding block; and
derive a prediction pixel value from the first and second predictors, wherein a plurality of prediction pixel values make up a prediction block.
2. The method of claim 1 , wherein deriving the prediction pixel value comprises averaging the first and second predictors.
3. The method of claim 1 , wherein the intensity value of the neighboring pixel from the row of horizontal boundary pixels on the upper side of the current coding block, used in calculating the intensity values for the neighboring pixels in the column along the right side of the current coding block, is a top right corner neighboring pixel.
4. The method of claim 1 , wherein the intensity value of the neighboring pixel from the column of vertical boundary pixels on the left side of the current coding block, used in calculating the intensity values for the neighboring pixels in the column along the left side of the current coding block, is a bottom left corner neighboring pixel.
5. The method of claim 1 , wherein calculating the intensity value of the bottom right neighboring pixel comprises calculating a weighted average of intensity values of two or more reconstructed neighboring pixels.
6. The method of claim 5 , wherein the intensity value of at least one of the neighboring pixels is filtered for use in the weighted average calculation of the intensity values.
7. The method of claim 5 , where R(W,−1) denotes an intensity value of a top right corner neighboring pixel at coordinates W,−1 and R(−1,H) denotes an intensity value of a bottom left corner neighboring pixel at coordinates −1,H, and the weighted average calculation is for the bottom right neighboring pixel P(W,H) and comprises as at least one of:
P ( W,H )=( W*R ( W,− 1)+ H*R (−1, H ))/( H+W ), or
P ( W,H )= R ( W,− 1)+ R (−1, H )− R (0,0).
8. The method of claim 7 , wherein the top right corner neighboring pixel is filtered before use in the weighted average calculation of the bottom right neighboring pixel according to at least one of:
R
filtered
(
W
,
-
1
)
=
(
R
(
W
-
3
,
-
1
)
+
3
*
R
(
W
-
2
,
-
1
)
+
5
*
R
(
W
-
1
,
-
1
)
+
7
*
R
(
W
,
-
1
)
)
16
R
filtered
(
W
,
-
1
)
=
(
R
(
W
-
2
,
-
1
)
+
2
*
R
(
W
-
1
,
-
1
)
+
5
*
R
(
W
,
-
1
)
)
8
.
9. The method of claim 7 , wherein the bottom left corner neighboring pixel is filtered before use in the weighted average calculation of the bottom right neighboring pixel according to at least one of:
R
filtered
(
-
1
,
H
)
=
(
R
(
-
1
,
H
-
3
)
+
3
*
R
(
-
1
,
H
-
2
)
+
5
*
R
(
-
1
,
H
-
1
)
+
7
*
R
(
-
1
,
H
)
)
16
R
filtered
(
-
1
,
H
)
=
(
R
(
-
1
,
H
-
2
)
+
2
*
R
(
-
1
,
H
-
1
)
+
5
*
R
(
-
1
,
H
)
)
8
.
10. The method of claim 1 , wherein the neighboring pixel used in the calculation of intensity values of the neighboring pixels along the lower side of the current coding block is a bottom left corner neighboring pixel R(−1,H) at coordinates x=−1 and y=H.
11. The method of claim 10 , wherein the intensity values of the neighboring pixels along the lower side of the current coding block is computed by
P
b
(
x
,
H
)
=
(
W
-
1
-
x
)
*
R
(
-
1
,
H
)
+
(
x
+
1
)
*
P
(
W
,
H
)
W
.
12. The method of claim 11 , wherein for each neighboring pixel R(m,H) . . . R(W−1,H) along the lower side of the current coding block where integer m<W, filter the respective intensity value for use in the second predictor calculation.
13. The method of claim 1 , wherein the neighboring pixel used in the calculation of the intensity values of the neighboring pixels in the column along the right side of the current coding block is a top right corner neighboring pixel R(W,−1) at coordinates x=W and y=−1 coordinates.
14. The method of claim 13 , wherein the intensity values for the neighboring pixels in a column along the right side of the current coding block are computed by
P
r
(
W
,
y
)
=
(
H
-
1
-
y
)
*
R
(
W
,
-
1
)
+
(
y
+
1
)
*
P
(
W
,
H
)
H
where R(W,−1) denotes an intensity value of the top right corner neighboring pixel at x=W and y=−1 coordinates.
15. The method of claim 14 , wherein for each neighboring pixel R(W,n) . . . R(W,H−1) in the column along the right side of the current coding block where integer n<H, filter the respective intensity value for use in the first predictor calculation.
16. The method of claim 14 , wherein the first predictor is a vertical predictor and is determined according to P h (x,y)=(W−1−x)*R(−1,y)+(x+1)*P r (W,y).
17. The method of claim 11 , wherein the second predictor is a horizontal predictor and is determined according to P v (x,y)=(H−1−y)*R(x,−1)+(y+1)*P b (x,H).
18. The method of claim 1 , where the top right neighboring pixel is at coordinates W−1,−1 and the bottom left neighboring pixel is at coordinates −1,H−1, wherein calculating the intensity value of the bottom right neighboring pixel P(W−1, H−1) comprises calculating a weighted average of intensity values of the top right neighboring pixel R(W−1,−1) and the bottom left neighboring pixel R(−1,H−1) according to P(W−1,H−1)=(W*R(W−1,−1)+H*R(−1,H−1))/(H+W).
19. The method of claim 1 , wherein the neighboring pixel used in the calculation of the intensity values of the neighboring pixels in the row along the lower side of the current coding block is the bottom left neighboring pixel R(−1,H−1), the intensity value of a neighboring pixel in the row along the lower side of the current coding block computed by
P
b
(
x
,
H
-
1
)
=
(
W
-
1
-
x
)
*
R
(
-
1
,
H
-
1
)
+
(
x
+
1
)
*
P
(
W
-
1
,
H
-
1
)
W
where R(−1,H−1) denotes an intensity value of the bottom left neighboring pixel at x=−1 and y=H−1 coordinates.
20. The method of claim 1 , wherein the neighboring pixel used in the calculation of the intensity values of the neighboring pixels in the column along the right side of the current coding block is the top right neighboring pixel R(W−1,−1) at coordinates x=W−1 and y=−1, and the intensity values of the neighboring pixels in the column along the right side of the current coding block is computed by
P
r
(
W
-
1
,
y
)
=
(
H
-
1
-
y
)
*
R
(
W
-
1
,
-
1
)
+
(
y
+
1
)
*
P
(
W
-
1
,
H
-
1
)
H
.
21. The method of claim 19 , wherein the first predictor is a horizontal predictor and is determined according to P h (x,y)=(W−1−x)*R(−1,y)+(x+1)*P r (W−1,y).
22. The method of claim 20 , wherein the second predictor is a vertical predictor and is determined according to P v (x,y)=(H−1−y)*R(x,−1)+(y+1)*P b (x,H−1).