Template-matching based adaptive motion vector resolution by using an adaptive order of motion vector resolutions

A method for video encoding, comprising:

• determining a first motion vector predictor (MVP) corresponding to a first motion vector resolution (MVR) of ¼-pel based on motion information and the first MVR, a first plurality of MVRs that corresponds to a first reference picture including the first MVR that is a highest spatial resolution of the first plurality of MVRs and one or more remaining MVRs of the first plurality of MVRs that are lower spatial resolutions than ¼-pel, each of the first plurality of MVRs corresponding to a respective one of first MVPs that include the first MVP;
• determining each remaining MVP in the first MVPs for each remaining MVR in the first plurality of MVRs based on the determined first MVP and the respective remaining MVR in the first plurality of MVRs;
• performing template matching (TM) by determining TM costs based at least on the first plurality of MVRs and the first MVPs;
• determining an order of the first plurality of MVRs based on the TM costs;
• encoding a current block in a current picture based on one of the first plurality of MVRs, an adaptive motion vector prediction (AMVP) mode being applied to the current block with an adaptive motion vector resolution (AMVR) mode;
• and
• encoding a signaled MVR index that indicates the one of the first plurality of MVRs according to the order of the first plurality of MVRs.

A non-transitory computer-readable storage medium storing instructions which when executed by a processor cause the processor to perform an encoding method comprising:

• determining a first motion vector predictor (MVP) corresponding to a first motion vector resolution (MVR) of ¼-pel based on motion information and the first MVR, a first plurality of MVRs that corresponds to a first reference picture including the first MVR that is a highest spatial resolution of the first plurality of MVRs and one or more remaining MVRs of the first plurality of MVRs that are lower spatial resolutions than ¼-pel, each of the first plurality of MVRs corresponding to a respective one of first MVPs that include the first MVP;
• determining each remaining MVP in the first MVPs for each remaining MVR in the first plurality of MVRs based on the determined first MVP and the respective remaining MVR in the first plurality of MVRs;
• performing template matching (TM) by determining TM costs based at least on the first plurality of MVRs and the first MVPs;
• determining an order of the first plurality of MVRs based on the TM costs;
• encoding, in a bitstream, a current block in a current picture based on one of the first plurality of MVRs, an adaptive motion vector prediction (AMVP) mode being applied to the current block with an adaptive motion vector resolution (AMVR) mode;
• encoding, in the bitstream, a signaled MVR index that indicates the one of the first plurality of MVRs according to the order of the first plurality of MVRs;
• and
• transmitting the bitstream.