Video Decoding Chroma Prediction for Reduced Small-Block Delay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In video coding schemes using dual trees for luma and chroma, processing delays are increased due to the need to process chroma images after luma images, and small blocks require significant overhead, leading to inefficient throughput and parallel processing issues.
Innovation Solution
Implement a video decoding apparatus that uses a CT information decoder to derive specific flags and modes for chroma intra prediction, allowing for reduced processing delay by optimizing tree splits and intra prediction modes for small blocks, particularly using a dual tree structure for chroma.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If dual tree structure is used for luma and chroma, then coding flexibility is improved, but processing delay increases due to sequential processing requirement
Solution Approach 1:
The patent segments the chroma prediction process into multiple stages: first performing intra prediction on larger chroma blocks using luma data, then performing inter prediction on smaller chroma blocks using the already decoded chroma data from previous CTUs. This segmentation allows different processing strategies for different block sizes and types, improving overall processing efficiency while maintaining dual-tree flexibility.
Solution Approach 2:
The patent performs preliminary intra prediction on chroma blocks using luma prediction data before performing inter prediction. By pre-calculating intra prediction results and storing them, the system avoids redundant calculations and enables parallel processing of certain operations, thereby reducing overall processing delay while maintaining coding flexibility.
2Manufacturing precision
If small blocks are used for chroma coding, then coding precision is improved, but overhead per pixel increases and throughput deteriorates
Solution Approach 1:
The patent applies different prediction modes and block sizes locally based on the specific characteristics of each chroma block. Larger blocks use intra prediction for better precision, while smaller blocks use inter prediction to reduce overhead. This local adaptation optimizes the balance between coding precision and throughput by matching the prediction method to the block characteristics.
Solution Approach 2:
The patent dynamically changes prediction parameters such as block size, prediction mode (intra vs. inter), and reference block selection based on block dimensions and content characteristics. By adapting these parameters locally rather than using fixed settings, the system achieves high coding precision for small blocks while maintaining efficient throughput through optimized parameter selection.
3Measurement precision
If intra prediction with multiplication and oblique direction reference is used, then prediction accuracy is improved, but processing delay increases significantly
Solution Approach 1:
The patent applies complex intra prediction with multiplication and oblique direction reference only partially - specifically, only for certain block sizes and prediction modes where the accuracy benefit outweighs the processing cost. For other cases, simpler prediction methods are used, achieving a balance between prediction accuracy and processing delay by avoiding excessive computation where not strictly necessary.
Data Source
AI summary
In conventional cases that different trees are used between luma and chroma and that luma chroma prediction including multiplication in chroma intra prediction is used, there is a problem in that a processing delay caused by the chroma intra prediction is increased in small blocks. In a case that two different trees are used between luma and chroma, in a case of an 8×8 block in a luma size, 0 is derived without decoding qt_split_cu_flag, and in a case of an 8×4/4×8 block in a luma size, 0 is derived without decoding mtt_split_cu_flag. In a case that the size of a target block is smaller than a prescribed size, a chroma intra mode is derived from any one of DC, leftward prediction, and upward prediction.