Image Inter Prediction Using Reconstructed Pixel Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression standards like HEVC face inefficiencies in inter prediction, particularly when multiple motions are present in a block, leading to increased overhead and reduced encoding efficiency, and require transmission of motion vector information, which increases data volume.
Innovation Solution
The method involves deriving motion vectors from reconstructed pixel regions adjacent to the current block, allowing for efficient inter prediction without transmitting motion vector information, and partitioning blocks into sub-blocks based on neighboring block structures for improved encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vector information is transmitted from encoding device to decoding device, then inter prediction accuracy is improved, but data transmission volume increases
Solution Approach 1:
The decoding device autonomously derives motion vectors by performing motion estimation on reconstructed pixel regions from the current picture and corresponding regions from reference pictures, eliminating the need for the encoding device to transmit motion vector information. This self-service approach maintains prediction accuracy while reducing data transmission volume.
Solution Approach 2:
The decoding device performs motion estimation in advance using reconstructed pixel regions that are already available from previous decoding operations. By preparing motion vectors through preliminary motion estimation on locally available data, the system avoids needing to receive pre-computed motion vectors from the encoder.
2Measurement precision
If blocks are partitioned into small parts to handle multiple motions, then motion representation accuracy is improved, but encoding overhead increases
Solution Approach 1:
The current picture is divided into multiple reconstructed pixel regions, each of which can be independently used for motion estimation. This segmentation allows the system to handle multiple motions within a block by performing separate motion estimations on different regions, improving motion representation accuracy without requiring complex partitioning schemes.
Solution Approach 2:
The decoding device copies motion estimation results from reconstructed pixel regions directly for use in inter prediction. By reusing the motion estimation process and results from already-decoded regions, the system avoids the overhead of transmitting multiple motion vectors while still capturing multiple motion patterns.
3Productivity
If reconstructed pixel regions are used for motion estimation, then encoding efficiency is improved, but computational complexity at decoding side increases
Solution Approach 1:
The decoding device performs its own motion estimation using reconstructed pixel regions that are naturally available during the decoding process. This self-service approach leverages the fact that reconstructed regions must already be processed and stored in memory, making the additional motion estimation computationally efficient rather than burdensome.
Solution Approach 2:
The motion estimation process continues seamlessly using the same reconstructed pixel regions that are already being processed for display. By maintaining continuous use of these regions for both display preparation and motion estimation, the system avoids redundant processing and keeps computational complexity manageable.
Data Source
AI summary
The present invention relates to an image encoding/decoding method and device. In an image decoding method and device according to an embodiment of the present invention, a reconstructed pixel region within an image to which a current block to be decoded belongs is selected; a motion vector of the reconstructed pixel region is derived on the basis of the reconstructed pixel region and a reference image of the current block; and the derived motion vector is selected as a motion vector of the current block.