Global motion constrained motion vector in inter prediction
A decoder includes circuitry configured to receive a bitstream, extract a header associated with a current frame and including a signal characterizing that global motion is enabled and further characterizing parameters of a global motion model, and decoding the current frame, the decoding including using a motion model for each current block having a complexity that is less than or equal to a complexity of the global motion model. Related apparatus, systems, techniques, and articles are also described.
1. A decoder, the decoder comprising circuitry configured to:
receive a bitstream;
extract from the bitstream a parameter set associated with a coded picture in the bitstream, wherein the parameter set includes a signal indicating that a global motion model is enabled, the global motion model comprising an affine motion model using control point motion vectors; and
decode the coded picture using a motion model for each block of the coded picture, the motion model for each block having a complexity that is no greater than a complexity of the global motion model, wherein the complexity of a motion model is a function of a number of motion vectors needed to decode a block.
2. The decoder of claim 1 , wherein the parameter set includes a sequence parameter set (SPS).
3. The decoder of claim 1 , wherein the global motion model includes a four-parameter affine motion model.
4. The decoder of claim 1 , wherein the global motion model includes a six-parameter affine motion model.
5. The decoder of claim 1 , wherein each block is a coding tree unit.
6. The decoder of claim 1 , wherein each block is a coding unit.
7. The decoder of claim 1 , wherein each block in the coded picture is decoded using a translational motion model.
8. The decoder of claim 1 , wherein each block of a first region of the coded picture is decoded using the global motion model and each block in a second region of the coded picture is decoded using a translational motion model.
9. The decoder of claim 1 , wherein the global motion model is a six-parameter affine motion model, each block in a first region of the coded picture is decoded using the global motion model, and each blocks of a second region of the coded picture is decoded using one or both of a four-parameter affine motion model and a translational motion model.
10. A method comprising:
receiving, by a decoder, a bitstream;
extracting, by the decoder, from the bitstream a parameter set associated with a coded picture in the bitstream, wherein the parameter set includes a signal indicating that a global motion model is enabled, the global motion model comprising an affine motion model using control point motion vectors; and
decoding, by the decoder, the coded picture using a motion model for each block having a complexity that is no greater than a complexity of the global motion model, wherein the complexity of a motion model is a function of a number of motion vectors needed to decode a block.
11. The method of claim 10 , wherein the parameter set includes a sequence parameter set (SPS).
12. The method of claim 10 , wherein the global motion model includes a four-parameter affine motion model.
13. The method of claim 10 , wherein the global motion model includes a six-parameter affine motion model.
14. The method of claim 10 , wherein each block is a coding tree unit.
15. The method of claim 10 , wherein each block is a coding unit.
16. The method of claim 10 , wherein each block in the coded picture is decoded using a translational motion model.
17. The method of claim 10 , wherein each block of a first region of the coded picture is decoded using the global motion model and each block in a second region of the coded picture is decoded using a translational motion model.
18. The method of claim 10 , wherein the global motion model is a six-parameter affine motion model, each block in a first region of the coded picture is decoded using the global motion model, and each block of a second region of the coded picture is decoded using one or both of a four-parameter affine motion model and a translational motion model.