Method and apparatus for scalable motion vector coding

Abstract

A method and apparatus for scalable coding of a motion vector generated during motion estimation, in which a generated motion vector field is separated into a base layer and an enhancement layer according to pixel accuracies to obtain a layered structure for a motion vector. In addition, the motion vector field has a layered structure including a base layer composed of motion vectors of blocks larger than or equal to a predetermined size and at least one enhancement layer composed of motion vectors of blocks smaller than a predetermined size.

1 . A scalable motion vector coding method comprising:
 (a) dividing a current frame into a plurality of blocks and performing motion estimation to determine a motion vector for each of the divided blocks;   (b) forming a base layer including motion vectors of blocks larger than or equal to a predetermined size and at least one enhancement layer including motion vectors of blocks smaller than the predetermined size, using the motion vectors of the divided blocks; and   (c) coding the base layer and the enhancement layer.   
     
     
         2 . The scalable motion vector coding method of  claim 1 , wherein in (a), a motion vector for each of the divided blocks is determined using hierarchical variable size block matching. 
     
     
         3 . The scalable motion vector coding method of  claim 1 , wherein in (b), the base layer is formed by assigning a representative motion vector selected from motion vectors of adjacent small blocks smaller than the predetermined size among the divided blocks to a larger block formed by merging the small blocks and the enhancement layer is formed of the motion vectors of the merged small blocks. 
     
     
         4 . The scalable motion vector coding method of  claim 3 , wherein the representative motion vector is the first motion vector having a same type as the motion vector of the larger block among the motion vectors of the small blocks scanned according to a predetermined scan order. 
     
     
         5 . The scalable motion vector coding method of  claim 4 , wherein the predetermined scan order is a quad-tree scan order. 
     
     
         6 . The scalable motion vector coding method of  claim 4 , wherein the type of the motion vector is identified according to whether a reference frame indicated by the motion vector of the larger block is a previous or next frame of the current frame. 
     
     
         7 . The scalable motion vector coding method of  claim 1 , wherein (c) comprises:
 scanning blocks of the base layer and the enhancement layer according to a predetermined scan order and estimating the motion vectors of the base layer and the enhancement layer using motion vectors of neighboring blocks that are processed prior to the scanned blocks of the base layer and the enhancement layer; and   coding a difference between the motion vectors of the base layer and the enhancement layer, which are estimated using the motion vectors of the neighboring blocks, and actual motion vectors using context-based adaptive binary arithmetic coding.   
     
     
         8 . A scalable motion vector coder comprising:
 a motion estimation unit which divides a current frame into a plurality of blocks and performs motion estimation to determine a motion vector for each of the divided blocks;   a layer forming unit which forms a base layer including motion vectors of blocks larger than or equal to a predetermined size and at least one enhancement layer including motion vectors of blocks smaller than the predetermined size, using the motion vectors of the divided blocks; and   a layered motion vector coding unit which codes the base layer and the enhancement layer.   
     
     
         9 . The scalable motion vector coder of  claim 8 , wherein the motion estimation unit determines a motion vector for each of the divided blocks using hierarchical variable size block matching. 
     
     
         10 . The scalable motion vector coder of  claim 8 , wherein the layer forming unit forms the base layer by assigning a representative motion vector selected from motion vectors of adjacent small blocks smaller than the predetermined size among the divided blocks to a larger block formed by merging the small blocks and forms the enhancement layer with the motion vectors of the merged small blocks. 
     
     
         11 . The scalable motion vector coder of  claim 10 , wherein the representative motion vector is the first motion vector having a same type as the motion vector of the larger block among the motion vectors of the small blocks scanned according to a predetermined scan order. 
     
     
         12 . The scalable motion vector coder of  claim 11 , wherein the predetermined scan order is a quad-tree scan order. 
     
     
         13 . The scalable motion vector coder of  claim 11 , wherein the type of the motion vector is identified according to whether a reference frame indicated by the motion vector of the larger block is a previous or next frame of the current frame. 
     
     
         14 . The scalable motion vector coder of  claim 8 , wherein the layered motion vector coding unit scans blocks of the base layer and the enhancement layer according to a predetermined scan order, estimates the motion vectors of the base layer and the enhancement layer using motion vectors of neighboring blocks that are processed prior to the scanned blocks of the base layer and the enhancement layer, and codes a difference between the motion vectors of the base layer and the enhancement ‘layer, which are estimated using the motion vectors of the neighboring blocks, and actual motion vectors using context-based adaptive binary arithmetic coding. 
     
     
         15 . A scalable motion vector decoding method for decoding a motion vector coded into a base layer and an enhancement layer, comprising:
 extracting motion vector information by analyzing an input bitstream;   restoring motion vector components of the base layer and the enhancement layer from the extracted motion vector information; and   summing the restored motion vector components of the base layer and the enhancement layer to output a layered motion vector,   wherein the base layer includes information about motion vectors of blocks larger than or equal to a predetermined size and the enhancement layer includes information about motion vectors of blocks smaller than the predetermined size.   
     
     
         16 . A scalable motion vector decoder for decoding a motion vector coded into a base layer and an enhancement layer, the decoder comprising:
 a motion vector restoring unit which extracts motion vector information by analyzing an input bitstream and restoring motion vector components of the base layer and the enhancement layer from the extracted motion vector information; and   a summing unit which sums the restored motion vector components of the base layer and the enhancement layer to generate a layered motion vector,   wherein the base layer includes information about motion vectors of blocks larger than or equal to a predetermined size and the enhancement layer includes information about motion vectors of blocks smaller than the predetermined size.