Method for image encoding and apparatus for implementing the same using interpolation filter based on motion speed

Patent metadata

Field	Value
Publication number	US-11218696-B2
Application number	US-202016797300-A
Country	US
Kind code	B2
Filing date	Feb 21, 2020
Priority date	Feb 21, 2019
Publication date	Jan 4, 2022
Grant date	Jan 4, 2022

Abstract

A method of encoding a first image in a plurality of images of input video data, wherein the first image is divided into a plurality of pixel blocks comprises, for a current block of the first image: determining a motion speed for the current block with respect to a reference block correlated to the current block in a second image of the plurality of images, wherein the second image is distinct from the first image and was previously encoded according to an image encoding sequence for encoding the images of the plurality of images; determining an interpolation filter based on the motion speed; and using the interpolation filter for calculating fractional pixel values in the second image for a temporal prediction of pixels of the current block based on the reference block of the second image.

First claim

The invention claimed is:

1. A method of encoding a first image in a plurality of images of input video data, wherein the first image is divided into a plurality of pixel blocks, the method comprising, for a current block of the first image:
determining a motion speed for the current block with respect to a reference block correlated to the current block in a second image of the plurality of images, wherein the second image is distinct from the first image and was previously encoded according to an image encoding sequence for encoding the images of the plurality of images;
determining an interpolation filter based on the motion speed;
using the interpolation filter for calculating fractional pixel values in the second image for a temporal prediction of pixels of the current block based on the reference block of the second image,
determining a motion vector for the current block, the motion vector pointing to the reference block in the second image;
determining the motion speed based on the motion vector,

wherein the interpolation filter is determined by obtaining its transfer function as a product of a cardinal sine function and a window function,
wherein the argument of the cardinal sine function depends on the motion speed (m) and/or a width of a rectangular window function convolved in the time domain with the cardinal sine function depends on the motion speed (m), and
wherein the motion speed corresponds to a speed of motion of the current block relative to the reference block in the second image.

2. The method according to claim 1 , further comprising:
determining the motion speed based on an absolute value of the component of the motion vector in a direction of interpolation.

3. The method according to claim 1 , further comprising: determining the motion speed based on a temporal distance between the first and second images in the input video data.

4. The method according to claim 1 , wherein the interpolation filter transfer function g(i) is obtained as:

g
⁡

(
i
)

=

sin
⁢

⁢

c
⁡

(

π
⁡

(

i
+
p

)

D
⁡

(
m
)

)

·

w
⁡

(

i
+
p

)

wherein D(m) is obtained as:
D ( m )= j+k ·abs( {right arrow over (mv)} )

wherein abs({right arrow over (mv)}) is based on an absolute value of the component of the motion vector in a direction of interpolation, i is a discrete time value, p is a fractional phase offset which indicates a fractional pixel position, w is the window function, and wherein j and k are positive real numbers.

5. The method according to claim 1 , wherein the width of the rectangular window function convolved in the time domain with the cardinal sine function depends on the motion speed (m), and wherein the interpolation filter transfer function g(i) is obtained as:

g
⁡

(
i
)

=

(

sin
⁢

⁢

c
⁡

(

π
⁡

(

i
+
p

)

D
⁡

(
m
)

)

*
r
⁢
e
⁢
c
⁢

t
m

⁡

(
i
)

)

·

w
⁡

(

i
+
p

)

,

wherein i is a discrete time value, p is a fractional phase offset which indicates a fractional pixel position, w is the window function, * represents a discrete convolution operator, · is the regular multiplication, rect m (i) is the rectangular window function, and D is a real positive number.

6. The method according to claim 1 , wherein the argument of the cardinal sine function depends on the motion speed (m) and a width of a rectangular window function convolved in the time domain with the cardinal sine function depends on the motion speed (m), and wherein the interpolation filter transfer function g(i) is obtained as:

g
⁡

(
i
)

=

(

Assignees

• Ateme

Inventors

• Dorovic Katarina
• Mora Elie
• Monnier Mathieu

Classifications

• H04N19/573

  Motion compensation with multiple frame prediction using two or more reference frames in a given prediction direction · CPC title

• H04N19/117 Primary

  Filters, e.g. for pre-processing or post-processing (sub-band filter banks H04N19/635) · CPC title

• H04N19/139

  Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability · CPC title

• H04N19/523

  with sub-pixel accuracy · CPC title

• H04N19/59

  involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution · CPC title