Reference Frame Clustering for More Accurate Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies, such as HEVC, struggle with selecting optimal reference frames, leading to suboptimal motion estimation results due to the concentration of reference frames close to the to-be-coded frame, limiting bandwidth efficiency and coding accuracy.
Innovation Solution
A method for reference frame selection that involves determining attribute information of a current frame, selecting candidate reference frames based on this information, clustering them into forward and backward clusters, and choosing the most suitable frames for the reference frame list using clustering algorithms like K-means, ensuring diverse and accurate frame selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference frames are selected from the reference frame set using conventional methods, then the selection process is simple, but the accuracy of reference frame selection deteriorates due to concentration of reference frames close to the to-be-coded frame
Solution Approach 1:
The reference frame set is divided into multiple clusters based on temporal distance and similarity characteristics. By segmenting the reference frames into different groups (clusters), the method ensures diverse selection of reference frames with varying temporal distances, preventing concentration on frames too close to the current frame and improving motion estimation accuracy.
Solution Approach 2:
The method introduces a new dimension of diversity by clustering reference frames based on both temporal distance and similarity metrics. This multi-dimensional approach selects reference frames that are not only temporally distributed but also vary in content similarity, thereby improving the quality of motion estimation beyond what single-criterion selection can achieve.
2Manufacturing precision
If multiple candidate reference frames are selected for motion estimation, then the coding accuracy improves, but the transmission bandwidth and storage costs increase
Solution Approach 1:
The method dynamically adjusts the number of reference frames selected from each cluster based on coding performance requirements. By changing the selection parameters adaptively, the system achieves optimal coding accuracy while controlling the quantity of reference frames used, thereby balancing coding quality with bandwidth and storage efficiency.
3Productivity
If reference frames close to the current frame are selected, then the motion estimation is computationally simpler, but the coding performance deteriorates due to lack of temporal diversity
Solution Approach 1:
The reference frame set is segmented into multiple clusters with different temporal characteristics. This segmentation enables the selection of reference frames from various temporal distances, ensuring both computational feasibility and coding performance by avoiding over-concentration on temporally close frames while maintaining manageable complexity.
Solution Approach 2:
Different clusters are formed based on local characteristics of reference frames (temporal distance and similarity). By applying local quality principles, the method selects reference frames with appropriate temporal diversity for different coding scenarios, optimizing the balance between computational efficiency and coding performance for each specific situation.
Data Source
AI summary
A method for reference frame selection, an apparatus for reference frame selection, an electronic device and a storage medium are provided, which relates to the field of data processing technology, in particular to the fields of video coding technology and unsupervised learning technology. The method includes: acquiring a current frame to be processed and determining attribute information of the current frame; selecting candidate reference frames from a reference frame set according to the attribute information; clustering the selected candidate reference frames to obtain at least one cluster; and selecting one candidate reference frame from each of the at least one cluster and adding the one selected candidate reference frame from each of the at least one cluster to a reference frame list associated with the current frame. The technical solution herein provided can improve the accuracy of reference frame selection.