Block Vector Magnitude Prediction With Candidate Pruning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies face challenges in efficiently compressing video data due to the limitations in predicting and signaling block vector differences, leading to high overhead and reduced compression efficiency.
Innovation Solution
The use of a block vector difference (BVD) predictor to select a subset of candidates based on a cost function, predicting magnitude symbols of BVD, and signaling whether the prediction is correct, thereby improving compression efficiency by reducing the overhead needed to signal magnitude symbol predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If magnitude symbols of BVD are directly signaled without prediction, then signaling is simple, but compression efficiency is reduced due to uniform probability distribution
Solution Approach 1:
The patent applies preliminary action by predicting magnitude symbols of BVD before signaling. A BVD predictor is used to generate predicted magnitude symbols based on previously decoded blocks, and only the difference between actual and predicted values needs to be signaled. This preliminary prediction step transforms the uniform probability distribution into a non-uniform one, improving compression efficiency while maintaining signaling simplicity through selective signaling of only necessary correction data.
2Measurement precision
If all BVD candidates are used for prediction, then prediction accuracy may be maintained, but signaling overhead increases
Solution Approach 1:
The patent applies the taking out principle by extracting and selecting only the most relevant BVD candidates from the full candidate set. A subset of BVD candidates is identified based on prediction accuracy metrics, and only this reduced subset is used for magnitude symbol prediction. This extraction process removes unnecessary candidates that would increase signaling overhead, while maintaining prediction accuracy by retaining the most informative candidates.
Solution Approach 2:
The patent applies segmentation by dividing the BVD candidate set into a subset for prediction and excluding others. The candidates are segmented based on their relevance and prediction performance, creating a structured approach where only essential candidates participate in the prediction process. This segmentation reduces the complexity and signaling overhead associated with handling all candidates uniformly.
3Loss of information
If BVD candidates are pruned to a subset, then signaling overhead is reduced, but prediction accuracy may deteriorate
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the BVD candidate subset based on prediction performance parameters. The selection of which candidates to include in the subset is not fixed but adapts based on measured prediction accuracy and statistical properties of the video content. This parameter-based selection ensures that the reduced subset maintains optimal prediction accuracy while achieving signaling overhead reduction.
Data Source
AI summary
Encoding and/or decoding a block of a video frame may be based on a previously decoded reference block in the same frame or a different frame. The reference block may be indicated by a block vector (BV). A block vector difference (BVD) predictor may be used to make predictions about the magnitude of a BVD. The BVD predictor may be selected from BVD candidates, which may be pruned based on costs associated with them. Pruning the BVD candidates may improve additional predictions and thus improve the compression efficiency of the predictions. A motion vector difference (MVD) predictor may be used to make predictions about the magnitude of an MVD. The MVD may be used with a motion vector predictor (MVP) to indicate a motion vector (MV).