Video Codec Transform Optimization via Iterative Mode Pairing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video codecs face challenges in optimizing transforms and predictions, leading to suboptimal compression efficiency and increased processing time, especially in real-time environments with limited resources.
Innovation Solution
A joint optimization iterative algorithm that determines optimized mode pairs by iteratively modifying intra-predictors and transforms based on training videos and quantization parameters, accumulating block statistics to select the best mode pairs for each block, and repeating the process until performance meets a predetermined threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video codecs use fixed transforms and predictions, then implementation is simple, but compression efficiency is suboptimal
Solution Approach 1:
The patent applies dynamics by making the transform and prediction parameters adaptive rather than fixed. The system dynamically selects different transforms (e.g., DCT, ADCT, HDCT) and prediction modes based on block characteristics and coding context, allowing the algorithm to adjust its behavior to optimize compression efficiency for each specific situation rather than using a one-size-fits-all approach
Solution Approach 2:
The patent changes parameters by varying transform types, prediction modes, and coding configurations based on accumulated statistics and performance metrics. The system modifies parameters such as transform selection, prediction strength, and coding mode based on learned patterns from training data, enabling continuous optimization of compression efficiency without requiring complete algorithm redesign
2Productivity
If iterative optimization is performed to improve compression efficiency, then rate-distortion performance improves, but processing time increases
Solution Approach 1:
The patent applies preliminary action by performing iterative optimization during an offline training phase using training videos. The system pre-computes accumulated statistics and optimized parameters before actual video encoding occurs. This preliminary computation allows the online encoding process to use pre-optimized transforms and predictions, achieving high rate-distortion performance without the computational burden of real-time iterative optimization
Solution Approach 2:
The patent uses feedback by measuring rate-distortion performance during training and using these measurements to guide parameter optimization. The system accumulates statistics from encoding/decoding operations and uses this feedback to adjust transform selections and prediction modes. This feedback mechanism enables continuous improvement of compression efficiency while controlling processing time through data-driven decision making
3Manufacturing precision
If multiple transforms and predictions are tested to find optimal mode pairs, then compression quality improves, but computational resources increase
Solution Approach 1:
The patent applies local quality by selecting different transforms and predictions tailored to specific block characteristics rather than using a uniform approach. The system analyzes local block properties (e.g., texture, edge density, motion characteristics) and applies the most suitable transform and prediction mode for each local region. This localized optimization improves compression quality for each block while reducing overall computational resources by avoiding unnecessary transform applications
Solution Approach 2:
The patent uses copying by creating a library of pre-computed transform and prediction mode pairs from training data. Instead of exhaustively testing all possible transforms and predictions during actual encoding, the system copies and reuses optimized mode pairs that have been pre-identified through training. This copying approach maintains high compression quality while significantly reducing computational resources by replacing exhaustive search with selective retrieval from pre-computed options
Data Source
AI summary
A joint optimization iterative algorithm determines optimized mode pairs. Each mode pair includes an intra-predictor and a transform pair that are iteratively modified to determine an optimized intra-predictor and an optimized transform that forms the optimized mode pair. A set of training videos and a set of quantization parameters (QPs) are used as the base data for determining the optimized mode pairs. Each video includes a plurality of pixel blocks, herein referred to as blocks. Block statistics associated with each mode pair are accumulated by separately encoding each block using each mode pair, and selecting the best mode pair for each block according to a measured characteristic of each encoding. The accumulated block statistics are used to modify the intra-predictor and the transform within each mode pair.