Method and apparatus for video coding using motion vector with component-wise adaptive spatial resolution
AssigneesHyundai Motor Company, Kia Corporation, Research & Business Foundation Sungkyunkwan University
InventorsByeung Woo Jeon, Bum Yoon Kim, Yu Jin Lee, Seung Wook Park
Life of this patent
Estimating the term…
Abstract
A video coding method and apparatus adaptively determine spatial resolution for a horizontal component and a vertical component of a motion vector or block vector indicating a position of a reference block when performing inter prediction or intra block copy (IBC) of a current block. The video coding method and apparatus separately determine each spatial resolution for a plurality of motion vectors.
Claims11
A method of determining spatial resolution values of a first component and a second component with respect to a difference of a motion vector of a current block, performed by a video decoding apparatus, the method comprising:
- decoding a same resolution flag and a first component AMVR (Adaptive Motion Vector Resolution) flag from a bitstream, wherein the same resolution flag indicates whether the spatial resolution values of the first component and the second component are the same, and wherein the first component AMVR flag indicates whether or not adaptive spatial resolution is applied to the first component;
- in response to the same resolution flag being true:
- determining the spatial resolution values of the first component and the second component to be a predetermined same value in response to the first component AMVR flag being false, and
- determining the spatial resolution values of the first component and the second component to be a same value based on a first component precision index decoded from the bitstream in response to the first component AMVR flag being true;
- and
- in response to the same resolution flag being false:
- determining the spatial resolution values of the first component and the second component to be different from each other,
- wherein one of the first component and the second component indicates a horizontal component, and the other indicates a vertical component.
A method of encoding spatial resolution values of a first component and a second component with respect to a difference of a motion vector of a current block, performed by a video encoding apparatus, the method comprising:
- determining the spatial resolution values of the first component and the second component;
- and
- encoding a same resolution flag and a first component Adaptive Motion Vector Resolution (AMVR) flag into a bitstream based on the spatial resolution values of the first component and the second component,
- wherein encoding the same resolution flag and the first component AMVR flag comprises:
- determining that the same resolution flag is true and the first component AMVR flag is false in response to the spatial resolution values of the first component and the second component being equal to a predetermined value;
- determining that the same resolution flag is true and the first component AMVR flag is true in response to the spatial resolution values of the first component and the second component being the same value that is different from the predetermined value;
- encoding a first component precision index indicating the spatial resolution values of the first component and the second component in response to the spatial resolution values of the first component and the second component being the same value that is different from the predetermined value;
- and
- determining that the same resolution flag is false in response to the spatial resolution values of the first component and the second component being different, and
- wherein one of the first component and the second component indicates a horizontal component, and the other indicates a vertical component.
A method for transmitting a bitstream containing encoded video data, the method comprising:
- generating the bitstream for a current block in an image;
- and
- transmitting the bitstream,
- wherein generating the bitstream for the image comprises:
- determining spatial resolution values of a first component and a second component with respect to a difference of a motion vector of the current block, and
- encoding a same resolution flag and a first component AMVR Adaptive Motion Vector Resolution (AMVR) flag into the bitstream based on the spatial resolution values of the first component and the second component,
- wherein encoding the same resolution flag and the first component AMVR flag comprises:
- determining that the same resolution flag is true and the first component AMVR flag is false in response to the spatial resolution values of the first component and the second component equal to a predetermined value,
- determining that the same resolution flag is true and the first component AMVR flag is true in response to the spatial resolution values of the first component and the second component being the same value that is different from the predetermined value,
- encoding a first component precision index indicating the spatial resolution values of the first component and the second component in response to the spatial resolution values of the first component and the second component being the same value that is different from the predetermined value, and
- determining that the same resolution flag is false in response to the spatial resolution values of the first component and the second component being different, and
- wherein one of the first component and the second component indicates a horizontal component, and the other indicates a vertical component.
Dependencies are inferred from the claim text.