Image Coding With Motion-Vector Merging for Stationary Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding methods using inter prediction lack efficiency in accurately predicting motion vectors, particularly for stationary regions, leading to decreased coding efficiency.
Innovation Solution
An image coding method that derives merging candidates based on prediction direction, motion vector, and reference picture index from neighboring blocks, and introduces a predetermined vector as a second merging candidate, enhancing the variety and accuracy of motion vector prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion estimation is performed using conventional inter prediction methods, then coding is achieved, but coding efficiency is insufficient particularly for stationary regions
Solution Approach 1:
The patent divides the image into multiple blocks and further into coding units, allowing different prediction methods to be applied to different regions. Each block can independently select between inter prediction and merging mode, enabling optimized coding efficiency for stationary regions while maintaining overall coding performance.
Solution Approach 2:
The patent introduces a dynamic selection mechanism where the encoder can switch between inter prediction and merging mode based on the characteristics of each block. This dynamic adaptation allows the system to optimize coding efficiency by selecting the appropriate prediction method for each region, improving overall coding performance.
2Productivity
If merging mode is used to improve coding efficiency, then compression performance increases, but the complexity of determining optimal merging candidates increases
Solution Approach 1:
The patent pre-establishes a list of merging candidates based on neighboring blocks before performing the actual merging operation. This preliminary preparation reduces the complexity of determining optimal merging candidates during encoding, as the system only needs to select from pre-computed candidates rather than evaluating all possible combinations.
Solution Approach 2:
The patent uses copying of motion vectors and prediction parameters from neighboring blocks to generate merging candidates. This copying approach simplifies the complexity by reusing existing data rather than computing new values, thereby improving compression performance while maintaining manageable computational complexity.
Data Source
AI summary
A method for transmitting a bitstream via a network is provided. The bitstream being generated by: deriving a first candidate having a first motion vector that has been used to code a first block; deriving a second candidate having a zero motion vector for direction 0 and a zero motion vector for direction 1, and a reference picture index value of zero for each direction 0 and direction 1; and deriving a third candidate having a zero motion vector for direction 0 and a zero motion vector for direction 1, and the reference picture index value being incremented by 1 for each direction 0 and direction 1. One candidate from a plurality of candidates, including the first, second, and third candidate is selected, and an index identifying the selected one candidate is coded.