Motion Vector Coding by Using Representation of Norm and Angle Components

摘要

Since many video coding methods use prediction between motion-compensated frames, the motion vector must be transmitted. A large amount of code is generated for DCT coefficients for the motion vector. Therefore, prediction is performed from the motion vectors of neighboring blocks and variable-length coding of the horizontal and vertical components of the prediction error vector is performed to reduce the amount of generated code. However, since this method handles the horizontal and vertical components as independent components, appropriate code assignment for coding the prediction error vector may not be possible. In this paper, the authors propose a method of coding by using norm and angle components to represent the prediction error vector and compare its coding efficiency with the conventional method to show its effectiveness.