Image Prediction Block Blending for Accurate Motion Vector Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The rapid increase in multimedia data demand outpaces the development of channel bandwidth, necessitating improved video compression techniques, particularly in predicting current blocks using multiple prediction blocks and enhancing motion information accuracy for efficient encoding and decoding.
Innovation Solution
An image decoding method that generates a first prediction block through inter prediction and a second block through intra prediction, combining them via weighted summation, and an encoding method that determines an initial motion vector using past and future directions to optimize motion vector search, incorporating DMVD mode operation information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple prediction blocks are used to predict current block, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent combines inter prediction block and intra prediction block through weighted summation to create a final prediction block. This merging of multiple prediction approaches achieves higher prediction accuracy while managing computational complexity through a unified processing framework.
Solution Approach 2:
The patent introduces weight parameters (first weight and second weight) that control the contribution of inter and intra prediction blocks. By adjusting these parameters, the system can optimize prediction accuracy while controlling computational resources based on different coding conditions.
2Reliability
If motion vector search is optimized using past and future directions, then inter prediction performance is improved, but encoding complexity increases
Solution Approach 1:
The patent divides motion vector derivation into separate past-direction and future-direction components. This segmentation allows independent optimization of each direction's motion search while maintaining overall inter prediction accuracy, managing encoding complexity through modular processing.
Solution Approach 2:
The patent performs preliminary motion vector estimation using past and future reference frames before final prediction. This preliminary action provides initial motion information that guides subsequent refinement steps, improving inter prediction performance while organizing computational work in efficient stages.
3Measurement precision
If weighted summation of prediction blocks is performed, then prediction accuracy is improved, but processing time increases
Solution Approach 1:
The patent uses weight parameters to control the blending of prediction blocks. These parameters can be optimized or pre-determined to reduce the computational burden of weighted summation while maintaining prediction accuracy, thus reducing processing time.
Solution Approach 2:
The patent applies weighted summation selectively rather than uniformly across all blocks. By applying the computationally intensive weighted blending only where needed (partial action), the system maintains prediction accuracy improvements while minimizing overall processing time increase.
Data Source
AI summary
Disclosed herein are an image encoding method and an image decoding method. The image decoding method includes determining an initial motion vector of a current block using a motion vector of a reconstructed region, searching for the motion vector of the current block based on the initial motion vector, and generating a prediction sample of the current block using the motion vector. The initial motion vector includes a motion vector in a past direction and a motion vector in a future direction.