Image decoding method and apparatus based on inter-prediction in image coding system

1. A video decoding method performed by a decoding apparatus, comprising:

obtaining information for inter prediction of a current block from a bitstream;

constructing a motion vector predictor (MVP) candidate list based on a neighboring block of the current block;

deriving costs of MVP candidates included in the MVP candidate list;

deriving a modified MVP candidate list based on the costs of the MVP candidates;

deriving a MVP of the current block based on the modified MVP candidate list;

deriving a motion vector of the current block based on the MVP of the current block and a motion vector difference (MVD); and

performing prediction of the current block based on the motion vector and a reference picture index of the current block,

wherein the information for the inter prediction includes the MVD and the reference picture index,

wherein the cost of the MVP candidate is derived as the sum of a top cost of the MVP candidate and a left cost of the MVP candidate,

the top cost of the MVP candidate is derived based on a top luma cost of the MVP candidate, a top chroma cost of the MVP candidate, a width of a luma block of the current block, a width of a chroma block of the current block and a weighting factor,

the left cost of the MVP candidate is derived based on a left luma cost of the MVP candidate, a left chroma cost of the MVP candidate, a height of the luma block of the current block, a height of the chroma block of the current block and the weighting factor.

2. The video decoding method of claim 1 , wherein the neighboring block includes a left neighboring block of the current block, a top neighboring block of the current block, a top-right neighboring block of the current block, a bottom-left neighboring block of the current block, and a top-left neighboring block of the current block.

3. The video decoding method of claim 2 , wherein the constructing of the MVP candidate list based on the neighboring block of the current block includes

determining availabilities in the order of the motion vector of the left neighboring block of the current block, the motion vector of the top neighboring block, the motion vector of the top-right neighboring block, the motion vector of the bottom-left neighboring block, and the motion vector of the top-left neighboring block, and

sequentially adding available motion vectors to the MVP candidate list as the MVP candidate.

4. The video decoding method of claim 3 , wherein the reference picture index indicates a specific reference picture included in a specific reference picture list,

it is sequentially determined whether the motion vector of the neighboring block corresponds to a first motion vector and a second motion vector, and

when the motion vector of the neighboring block is not the first motion vector and the second motion vector, it is determined that the motion vector of the neighboring block is not available, and

wherein the first motion vector is the motion vector for the specific reference picture included in the specific reference picture list, and

the second motion vector is the motion vector for the specific reference picture included in a reference picture list other than the specific reference picture list.

5. The video decoding method of claim 2 , wherein the maximum number of MVP candidates included in the MVP candidate list is 5.

6. The video decoding method of claim 1 , wherein the top luma cost of the MVP candidate is derived as a sum of absolute difference (SAD) of a top template of the luma block of the current block and thea top template of a luma block of a reference block for the MVP candidate, and

the top chroma cost of the MVP candidate is derived as the SAD of a top template of the chroma block of the current block and a top template of a chroma block of the reference block for the MVP candidate.

7. The video decoding method of claim 6 , wherein the top cost of the MVP candidate is derived based on the following equation,

costA( LX )=( w *costA( LX,L )>>log 2( W L ×2))+((1− w )*costA( LX,C )>>log 2( W C ))

here, costA(LX) represents the top cost of the MVP candidate, costA(LX, L) represents the top cost of the luma block of the current block, costA(LX, C) represents the top cost of the chroma block of the current block, W L represents the width of the luma block of the current block, W C represents the width of the chroma block of the current block, and w represents the weighting factor.

8. The video decoding method of claim 1 , wherein the left luma cost of the MVP candidate is derived as a sum of absolute difference (SAD) of a left template of the luma block of the current block and a left template of a luma block of a reference block for the MVP candidate, and

the left chroma cost of the MVP candidate is derived as the SAD of a left template of the chroma block of the current block and a left template of a chroma block of the reference block for the MVP candidate.

9. The video decoding method of claim 8 , wherein the left cost of the MVP candidate is derived based on the following equation,

costL( LX )=( w *costL( LX,L )>>log 2(2× H L ))+((1− w )*costL( LX,C )>>log 2( H C ))

here, costL(LX) represents the left cost of the MVP candidate, costL(LX, L) represents the left cost of the luma block of the current block, costL(LX, C) represents the left cost of the chroma block of the current block, H L represents the height of the luma block of the current block, H C represents the height of the chroma block of the current block, and w represents the weighting factor.

10. The video decoding method of claim 1 , wherein the deriving of the modified MVP candidate list based on the costs for the MVP candidates includes

deriving a reordered MVP candidate list by reordering the MVP candidates in the order of small cost,

deriving a refine MVP candidate based on a specific MVP candidate among the MVP candidates, and

deriving the modified MVP candidate list by adding the refine MVP candidate to the reordered MVP candidate list.

11. A decoding apparatus for performing video decoding, comprising:

an entropy decoder to obtain information for inter prediction of a current block from a bitstream; and

a predictor to construct a motion vector predictor (MVP) candidate list based on a neighboring block of the current block, derive costs of MVP candidates included in the MVP candidate list, derive a modified MVP candidate list based on the costs of the MVP candidates, derive an MVP of the current block based on the modified MVP candidate list, derive a motion vector of the current block based on the MVP of the current block and a motion vector difference (MVD), and perform prediction of the current block based on a reference picture index and the motion vector of the current block,

wherein the information for the inter prediction includes the reference picture index and the MVD,

wherein the cost of the MVP candidate is derived as the sum of a top cost of the MVP candidate and a left cost of the MVP candidate,

the top cost of the MVP candidate is derived based on a top luma cost of the MVP candidate, a top chroma cost of the MVP candidate, a width of a luma block of the current block, a width of a chroma block of the current block and a weighting factor,

the left cost of the MVP candidate is derived based on a left luma cost of the MVP candidate, a left chroma cost of the MVP candidate, a height of the luma block of the current block, a height of the chroma block of the current block and the weighting factor.

12. An encoding apparatus for performing video encoding, comprising:

a predictor to construct a motion vector predictor (MVP) candidate list based on a neighboring block of a current block, derive costs of MVP candidates included in the MVP candidate list, derive a modified MVP candidate list based on the costs of the MVP candidates, derive an MVP of the current block based on the modified MVP candidate list, derive a motion vector of the current block based on the MVP of the current block, and perform prediction of the current block based on the motion vector of the current block; and

an entropy encoder to encode information for inter prediction of the current block,

wherein the information for the inter prediction includes a reference picture index indicating a reference picture of the current block and a motion vector difference (MVD),

wherein the cost of the MVP candidate is derived as the sum of a top cost of the MVP candidate and a left cost of the MVP candidate,

the top cost of the MVP candidate is derived based on a top luma cost of the MVP candidate, a top chroma cost of the MVP candidate, a width of a luma block of the current block, a width of a chroma block of the current block and a weighting factor,

the left cost of the MVP candidate is derived based on a left luma cost of the MVP candidate, a left chroma cost of the MVP candidate, a height of the luma block of the current block, a height of the chroma block of the current block and the weighting factor.