Video Coding Reference Boundary Correction for Prediction Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing picture coding and decoding techniques suffer from poor prediction efficiency due to the limited use of decoded samples in the neighbor of the coding/decoding target block for prediction.
Innovation Solution
A device is provided that includes a block vector candidate derivation unit, a selector, a storage for coded pictures of intra block copy standard blocks, and a reference region boundary correction unit to enhance prediction efficiency by considering a broader referenceable region.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only decoded samples in the neighbor of the coding/decoding target block are used for prediction, then the prediction process is simple, but the prediction efficiency is poor
Solution Approach 1:
The patent extends the reference region from the immediate neighbor blocks to a broader area including blocks from previously decoded pictures. This dimensional expansion in the reference space allows accessing more prediction candidates while maintaining systematic organization through picture order count (POC) based grouping and block vector management.
Solution Approach 2:
The patent performs preliminary organization of reference blocks by storing them in the reference block storage unit with their POC values and block vectors recorded in advance. This preliminary action enables efficient retrieval and selection of appropriate reference blocks during the prediction process without increasing computational complexity.
2Reliability
If a broader referenceable region is considered for prediction, then prediction efficiency is improved, but processing load increases
Solution Approach 1:
The patent segments the broader reference region into organized groups based on picture order count (POC) values. Reference blocks are divided into first reference blocks (POC=0) and second reference blocks (POC>0), with further segmentation by block vectors. This segmentation reduces the search space and enables efficient retrieval without processing the entire reference region.
Solution Approach 2:
The patent applies local quality by selecting reference blocks based on their specific properties (POC value, block vector, and spatial position) rather than uniformly processing all reference blocks. The reference region boundary correction unit selectively identifies and corrects boundaries for relevant blocks, focusing computational effort on locally optimal reference candidates.
3Measurement precision
If reference blocks from multiple pictures are utilized, then prediction accuracy is enhanced, but reference region management becomes complex
Solution Approach 1:
The patent introduces the reference block storage unit as an intermediary structure that manages reference blocks from multiple pictures. This intermediary organizes blocks by POC values and block vectors, acting as a buffer between the complex multi-picture reference region and the simpler prediction process, thereby reducing management complexity.
Solution Approach 2:
The patent performs preliminary organization of reference blocks by storing them with their metadata (POC values, block vectors, positions) in the reference block storage unit before the actual prediction process. This preliminary action pre-processes the complex reference region management, enabling simple retrieval and selection during prediction without real-time complexity.
Data Source
AI summary
A picture coding device includes: a block vector candidate derivation unit that derives block vector candidates of a coding target block in a coding target picture from coding information stored in a coding information storage memory; a selector that selects a selected block vector from the block vector candidates; a storage that stores coded pictures of a predetermined number of intra block copy standard blocks immediately before the coding target block; and a reference region boundary correction unit that removes a coded picture of one intra block copy standard block in the storage from a referenceable region after completion of a coding process of the coding target block, and determines whether an upper left position and a lower right position of a reference block indicated by the selected block vector are both included in the referenceable region.