Motion Vector Coding Apparatus for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion vector coding methods in HEVC, such as AMVP, fail to improve coding efficiency due to the need for both interlayer and intra-layer reference motion vectors, leading to increased complexity and inefficiency.
Innovation Solution
A motion vector coding apparatus that integrates interlayer predicted motion vectors and neighboring reference motion vectors to simplify the coding process, allowing for equal handling of both types of references and reducing the need for additional flags and codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AMVP-based predicted vector index code and interlayer motion vector prediction flag are both used, then motion vector prediction capability is improved, but device complexity increases
Solution Approach 1:
The patent merges the AMVP-based predicted vector index code mechanism with the interlayer motion vector prediction flag mechanism into a unified motion vector prediction system. The motion vector prediction unit integrates both reference types (intra-layer neighboring blocks and inter-layer base layer blocks) and selects the most appropriate predicted motion vector from combined reference sets, eliminating the need for separate processing paths and reducing overall system complexity while maintaining prediction capability.
Solution Approach 2:
The motion vector prediction unit is designed with multi-functionality to handle both AMVP prediction and interlayer prediction within a single unified framework. The same prediction unit can select from diverse reference sources (neighboring blocks in current layer, corresponding blocks in base layer) using a single selection mechanism, making the system more versatile and reducing the need for multiple specialized components.
2Productivity
If multiple reference motion vectors are used for prediction, then coding efficiency is improved, but the number of codes and flags increases
Solution Approach 1:
The patent combines multiple reference motion vector sources (intra-layer neighboring blocks and inter-layer base layer blocks) into a unified reference set. Instead of separately coding predictions from different reference types, the system integrates them into a single prediction process that selects the best match from the combined pool, reducing the total number of codes and flags needed while maintaining access to diverse reference information.
Data Source
AI summary
According to exemplary embodiments of the present invention, references to a motion vector in a neighboring block of an enhancement layer or in a block of a previously coded picture, and a motion vector of a base layer are equally handled. Accordingly, a motion_prediction_flag with identification information can be integrated for further improved coding efficiency.