Method and Device for Encoding/Decoding Image and Recording Medium Having Bitstream Stored Thereon

Abstract

The present invention relates to an image decoding method. The image decoding method comprises deriving initial motion information of a current block from at least one of motion information of a spatial neighbor block, motion information of a temporal neighbor block, and pre-defined motion information, generating refined motion information by performing motion information refinement for the initial motion information and generating a prediction block of the current block by using the refined motion information.

1 . An image decoding method performed by a decoding apparatus, the method comprising:
 deriving motion information of a current block;   deriving sub-blocks of the current block by partitioning the current block;   deriving refined motion information of each sub-block based on the motion information of the current block; and   deriving a prediction block of the current block based on the refined motion information of each sub-block.   
     
     
         2 . The method of  claim 1 , wherein the deriving the refined motion information of each sub-block comprises:
 deriving a first region for each sub-block of the current block in a L0 reference picture of the current block based on a L0 motion vector of the motion information of the current block;   deriving a second region for each sub-block of the current block in a L1 reference picture of the current block based on a L1 motion vector of the motion information of the current block;   deriving a Sum of Absolute Difference (SAD) for each sub-block of the current block based on the first region and the second region; and   deriving a refined L0 motion vector and a refined L1 motion vector of the refined motion information of each sub-block based on the SAD for each sub-block.   
     
     
         3 . The method of  claim 2 , wherein the refined L0 motion vector is derived in a search region in the L0 reference picture and the refined L1 motion vector is derived in a search region in the L1 reference picture. 
     
     
         4 . The method of  claim 3 , wherein the search region in the L0 reference picture of a sub pixel unit is restricted to be included in the search region in the L0 reference picture of an integer pixel unit, and
 wherein the search region in the L1 reference picture of a sub pixel unit is restricted to be included in the search region in the L1 reference picture of an integer pixel unit.   
     
     
         5 . The method of  claim 1 , wherein whether to derive the refined motion information is determined based on a difference between a Picture order count (POC) of the current picture and a POC of the L0 reference picture and a difference between a POC of the L1 reference picture and the POC of the current picture. 
     
     
         6 . The method of  claim 1 , wherein whether to derive the refined motion information is determined based on a size of the current block. 
     
     
         7 . An image encoding method performed by an encoding apparatus, the method comprising:
 deriving motion information of a current block;   deriving sub-blocks of the current block by partitioning the current block;   deriving refined motion information of each sub-block based on the motion information of the current block;   deriving a prediction block of the current block based on the refined motion information of each sub-block; and   encoding image information including prediction related information for the motion information of current block.   
     
     
         8 . The method of  claim 7 , wherein the deriving the refined motion information of each sub-block comprises:
 deriving a first region for each sub-block of the current block in a L0 reference picture of the current block based on a L0 motion vector of the motion information of the current block;   deriving a second region for each sub-block of the current block in a L1 reference picture of the current block based on a L1 motion vector of the motion information of the current block;   deriving a Sum of Absolute Difference (SAD) for each sub-block of the current block based on the first region and the second region; and   deriving a refined L0 motion vector and a refined L1 motion vector of the refined motion information of each sub-block based on the SAD for each sub-block.   
     
     
         9 . A non-transitory computer-readable storage medium storing a bitstream of encoded image information generated by the image encoding method of  claim 7 . 
     
     
         10 . A transmission method for image data, the method comprising:
 obtaining a bitstream of encoded image information, wherein the encoded image information is generated based on deriving motion information of a current block, deriving sub-blocks of the current block by partitioning the current block, deriving refined motion information of each sub-block based on the motion information of the current block, deriving a prediction block of the current block based on the refined motion information of each sub-block, and encoding the image information including prediction related information for the motion information of current block; and   transmitting the image data comprising the bitstream.