APPARATUS AND METHOD FOR ENCODING THREE-DIMENSIONAL MESH SEQUENCE THROUGH MOTION VECTOR ESTIMATION BY THE VERTEX UNIT

摘要

PURPOSE: An apparatus and a method for encoding a three-dimensional mesh sequence through motion vector estimation by the vertex unit are provided to easily predict a motion vector of each vertex of isomorphism mesh sequence. CONSTITUTION: A frame memory(102) updates and stores restored motion vectors of previous neighbor vertexes to a motion vector of an encoding target vertex. The first predictor(104) reads the motion vector of the previous neighbor vertexes and multiplies the optimum weight for predicting the motion vector of the encoding target vertex by the motion vectors of the neighbor vertexes for producing the first predicted motion vector. The second predictor(106) produces the second predicted motion vector of the optimum prediction mode having the lowest error rate from the second predicted candidate motion vectors. The second predicted candidate motion vectors are obtained by predicting the first error vector according to time or spatial prediction mode. A quantizer(112) quantizes the first error vector and the second error vector generated with a difference between the second predicted motion vectors. An arithmetic encoder(120) encodes the second quantized error vector data.