Intra prediction method and apparatus for video coding
A picture encoding method that includes selecting a V th basis function for a current block according to a specific order from among basis functions; deriving a V th expansion coefficient on the basis of the selected V th basis function; calculating a V th weighted error criterion on the basis of the selected V th basis function and the derived V th expansion coefficient; and encoding and outputting information indicating the V th number as a basis number when the value of the V th weighted error criterion is less than a specific critical value.
1. A method for encoding a picture performed by an encoding device, the method comprising:
deriving a V th basis function for encoding a current block according to a specific order among basis functions, wherein the basis functions are used for a frequency transform of a residual signal of the current block, and V is a positive integer;
deriving a V th extension coefficient for the V th basis function based on the V th basis function;
deriving a V th parametric model for the current block based on the V th basis function and the derived V th extension coefficient;
generating a V th residual error signal based on an original signal and the V th parametric model;
calculating a V th weighted error criterion based on the V th basis function and the V th residual error signal; and
based on a value of the V th weighted error criterion being smaller than a specific threshold value, encoding information representing V as a number of basis functions used in the V th parametric model,
wherein the parametric model is derived from following equation:
g
(
V
)
[
m
,
n
]
=
∑
(
k
,
l
)
∈
K
V
c
k
,
l
(
V
)
φ
k
,
l
[
m
,
n
]
herein, K v is a set of the basis functions selected from a first iteration to a V th iteration among the basis functions, (k,l) represents indexes of the basis functions selected from the first iteration to the V th iteration, c (v) k,l represents an extension coefficient for a basis function corresponding to the (k, l) index, φ k,l represents a basis function corresponding to the (k, l) index, and [m, n] represents a coordinate of a target sample in the current block.
2. The method for encoding a picture of claim 1 , further comprising:
deriving an intra-prediction mode for the current block,
wherein the basis functions are derived based on the intra-prediction mode.
3. The method for encoding a picture of claim 1 , further comprising:
deriving an intra-prediction mode for the current block,
wherein the specific order is determined based on the intra-prediction mode.
4. The method for encoding a picture of claim 1 , further comprising:
determining whether or not an intra-prediction based on the parametric model is used; and
encoding and outputting flag information representing whether or not the intra-prediction based on the V th parametric model is used based on the determination result.
5. The method for encoding a picture of claim 4 , wherein at least one of the intra-prediction based on the V th parametric model or the number of basis functions is used or determined, based on a size of the current block, respectively.
6. A method for decoding a picture performed by a decoding device, the method comprising:
obtaining information representing a number of basis functions used for a V th parametric model for a current block from a bitstream, wherein the basis functions are used for a frequency transform of a residual signal of the current block, and V is a positive integer;
deriving a V th basis function for a current block according to a specific order among the basis functions;
deriving V th extension coefficient for the V th basis function based on the V th basis function; and
checking whether V is equal to the number of basis functions; and
based on a V th number being equal to the number of basis functions, generating a prediction sample of the current block based on the V th basis function and the V th extension coefficient,
wherein the parametric model is generated from following equation:
g
(
V
)
[
m
,
n
]
=
∑
(
k
,
l
)
∈
K
V
c
k
,
l
(
V
)
φ
k
,
l
[
m
,
n
]
herein, K v is a set of the basis functions selected from a first iteration to a V th iteration among the basis functions, (k,l) represents indexes of the basis functions selected from the first iteration to the V th iteration, c (v) k,l represents an extension coefficient for a basis function corresponding to the (k, l) index, φ k,l represents a basis function corresponding to the (k, l) index, and [m, n] represents a coordinate of a target sample in the current block.
7. The method for decoding a picture of claim 6 , further comprising:
further obtaining information for an intra-prediction mode for the current block from the bitstream,
wherein the basis functions are derived based on the intra-prediction mode.
8. The method for decoding a picture of claim 6 , further comprising:
further obtaining information for an intra-prediction mode for the current block from the bitstream,
wherein the specific order is determined based on the intra-prediction mode.
9. The method for decoding a picture of claim 6 , further comprising:
obtaining flag information representing whether or not an intra-prediction based on the V th parametric model is used from the bitstream.
10. The method for decoding a picture of claim 9 ,
wherein the information representing the number of basis functions is obtained, based on a value of the flag information indicating 1, from the bitstream.
11. The method for decoding a picture of claim 9 , wherein at least one of the intra-prediction based on the V th parametric model or the number of basis functions is used or determined, based on a size of the current block, respectively.
12. The method for decoding a picture of claim 6 , wherein the number of basis functions is determined based on a quantization parameter (QP) value for the current block.