Video Block Prediction Using Parallel Interpolation Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current extrapolation filter prediction methods in video encoding and decoding exhibit low prediction efficiency, leading to poor video encoding and decoding performance.
Innovation Solution
Perform parallel prediction on at least two samples in a current block using an extrapolation filter, determining a reference region and filter coefficient to enhance prediction efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If sequential prediction is performed on samples in the current block using extrapolation filter prediction, then the prediction can be completed accurately, but the prediction efficiency is low resulting in poor video encoding and decoding performance
Solution Approach 1:
The current block is divided into multiple prediction blocks, and the prediction process is segmented into parallel operations. Each prediction block can be processed independently using the extrapolation filter with determined filter coefficients, allowing simultaneous prediction without interfering with each other's accuracy
Solution Approach 2:
The prediction process transitions from sequential single-sample processing to parallel multi-sample processing by adding the dimension of parallelism. Multiple samples in the current block are predicted simultaneously using the same extrapolation filter and filter coefficients, fundamentally changing the execution model from serial to parallel operations
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided in the present application are a video encoding method and apparatus, a video decoding method and apparatus, and a device, a system and a storage medium. The present application further discloses: when the current block is predicted, first determining a reference region of the current block, and an interpolation filter, determining a filtering coefficient on the basis of the reference region; on the basis of the filtering coefficient, performing parallel prediction on at least two pixel points in the current block by using the interpolation filter, so as to obtain a prediction block of the current block; and determining a transformation kernel corresponding to the current block, and determining a reconstruction value of the current block on the basis of the transformation kernel and the prediction block. That is to say, in the embodiments of the present application, when an interpolation filter is used for performing interpolation filtering prediction on the current block, parallel prediction is performed on at least two points in the current block, thereby improving the prediction speed, and as such improving the encoding and decoding efficiency.