Image encoding/decoding method and device for performing BDOF, and method for transmitting bitstream
An image encoding/decoding method and apparatus are provided. An image decoding method according to the present disclosure is performed by an image decoding apparatus. The image decoding method may comprise deriving a prediction sample of a current block based on motion information of the current block, determining whether bi-directional optical flow (BDOF) applies to the current block, based on that the BDOF applies to the current block, deriving a gradient for a current subblock in the current block, deriving motion refinement (v x , v y ) for the current subblock based on the gradient, deriving a BDOF offset based on the gradient and the motion refinement, and deriving a refined prediction sample for the current block based on the prediction sample of the current block and the BDOF offset.
1. An image decoding method performed by an image decoding apparatus, the image decoding method comprising:
deriving a L 0 prediction sample and a L 1 prediction sample of a current block by using motion information of the current block;
deriving a gradient for a predetermined area including a current subblock in the current block;
deriving motion offset (vx, vy) for the current subblock by using the gradient;
deriving a Bi-directional Optical Flow (BDOF) offset by using the gradient and the motion offset; and
deriving a prediction sample for the current block by using the L 0 prediction sample, the L 1 prediction sample of the current block and the BDOF offset,
wherein the gradient is derived by right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a first shift,
wherein the deriving the motion offset (vx, vy) comprises:
deriving a first intermediate parameter diff by subtracting the L 1 prediction sample of the current block from the L 0 prediction sample of the current block; and
deriving second intermediate parameters tempH and tempV by right-shifting the gradient by a third shift,
wherein the deriving the first intermediate parameter diff comprises right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a second shift,
wherein the first shift, the second shift and the third shift are set to a fixed value regardless of a bit depth of the current block,
wherein a value of the first shift is different from a value of the second shift and a value of the third shift, and
wherein a value of the second shift is different from a value of the third shift.
2. The image decoding method of claim 1 , wherein the first shift is 6.
3. The image decoding method of claim 1 , wherein the second shift is 4.
4. The image decoding method of claim 1 , wherein the third shift is 1.
5. The image decoding method of claim 1 , wherein the motion offset (vx, vy) is clipped in a predetermined range.
6. The image decoding method of claim 5 , wherein the predetermined range in which the motion offset (vx, vy) is clipped is set to a fixed range regardless of a bit depth of the current block.
7. The image decoding method of claim 1 ,
wherein the deriving the BDOF offset comprises right-shifting, by a predetermined shift, a value derived by using the gradient and the motion offset, and
wherein the predetermined shift is set to a fixed range regardless of a bit depth of the current block.
8. The image decoding method of claim 1 ,
wherein the deriving the prediction sample for the current block comprises clipping the BDOF offset in a predetermined range, and
wherein the predetermined range is set based on a bit depth of the current block.
9. An image decoding apparatus comprising:
a memory; and
at least one processor,
wherein the at least one processor is configured to:
derive a L 0 prediction sample and a L 1 prediction sample of a current block by using motion information of the current block;
derive a gradient for a predetermined area including a current subblock in the current block;
derive motion offset (vx, vy) for the current subblock by using the gradient;
derive a Bi-directional Optical Flow (BDOF) offset by using the gradient and the motion offset; and
derive a prediction sample for the current block by using the L 0 prediction sample, the L 1 prediction sample of the current block and the BDOF offset,
wherein the gradient is derived by right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a first shift,
wherein the deriving the motion offset (vx, vy) comprises:
deriving a first intermediate parameter diff by subtracting the L 1 prediction sample of the current block from the L 0 prediction sample of the current block; and
deriving second intermediate parameters tempH and tempV by right-shifting the gradient by a third shift,
wherein the deriving the first intermediate parameter diff comprises right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a second shift,
wherein the first shift, the second shift and the third shift are set to a fixed value regardless of a bit depth of the current block,
wherein a value of the first shift is different from a value of the second shift and a value of the third shift, and
wherein a value of the second shift is different from a value of the third shift.
10. An image encoding method performed by an image encoding apparatus, the image encoding method comprising:
deriving a L 0 prediction sample and a L 1 prediction sample of a current block by using motion information of the current block;
deriving a gradient for a predetermined area including a current subblock in the current block;
deriving motion offset (vx, vy) for the current subblock by using the gradient;
deriving a Bi-directional Optical Flow (BDOF) offset by using the gradient and the motion offset; and
deriving a prediction sample for the current block by using the L 0 prediction sample, the L 1 prediction sample of the current block and the BDOF offset,
wherein the gradient is derived by right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a first shift,
wherein the deriving the motion offset (vx, vy) comprises:
deriving a first intermediate parameter diff by subtracting the L 1 prediction sample of the current block from the L 0 prediction sample of the current block; and
deriving second intermediate parameters tempH and tempV by right-shifting the gradient by a third shift,
wherein the deriving the first intermediate parameter diff comprises right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a second shift,
wherein the first shift, the second shift and the third shift are set to a fixed value regardless of a bit depth of the current block,
wherein a value of the first shift is different from a value of the second shift and a value of the third shift, and
wherein a value of the second shift is different from a value of the third shift.
11. A method of transmitting a bitstream comprising:
transmitting a bitstream generated by an image encoding method,
wherein the image encoding method comprises,
deriving a L 0 prediction sample and a L 1 prediction sample of a current block by using motion information of the current block;
deriving a gradient for a predetermined area including a current subblock in the current block;
deriving motion offset (vx, vy) for the current subblock by using the gradient;
deriving a Bi-directional Optical Flow (BDOF) offset by using the gradient and the motion offset; and
deriving a prediction sample for the current block by using the L 0 prediction sample, the L 1 prediction sample of the current block and the BDOF offset,
wherein the gradient is derived by right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a first shift,
wherein the deriving the motion offset (vx, vy) comprises:
deriving a first intermediate parameter diff by subtracting the L 1 prediction sample of the current block from the L 0 prediction sample of the current block; and
deriving second intermediate parameters tempH and tempV by right-shifting the gradient by a third shift,
wherein the deriving the first intermediate parameter diff comprises right-shifting the L 0 prediction sample and the L 1 prediction sample of the current block by a second shift,
wherein the first shift, the second shift and the third shift are set to a fixed value regardless of a bit depth of the current block,
wherein a value of the first shift is different from a value of the second shift and a value of the third shift, and
wherein a value of the second shift is different from a value of the third shift.