Video Filter Information Prediction for Scalable Layered Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies face challenges in improving compression performance while reducing transmitted information amount, particularly in scalable and multi-view video coding/decoding scenarios.
Innovation Solution
Implementing filter information prediction methods that utilize information from other layers or views to generate prediction values for filtering, including adaptive loop, interpolation, and deblocking filters, and using indicators to manage filtering processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If filter information is transmitted for each layer or view in scalable and multi-view video coding, then filtering performance is improved, but transmitted information amount increases
Solution Approach 1:
The patent applies the copying principle by transmitting filter information from a reference layer or view and reusing it in the current layer or view. Instead of transmitting redundant filter information for each layer/view, the system copies the filter information from the reference and applies it after motion compensation, significantly reducing the transmitted information amount while maintaining filtering performance.
Solution Approach 2:
The patent implements universality by making the transmitted filter information applicable across multiple layers and views. A single set of filter information from a reference layer serves multiple purposes - it is used for filtering in the reference layer and also reused in enhancement layers and other views, making the information universally applicable and eliminating the need for separate transmissions.
2Measurement precision
If filter information is independently coded for each layer or view, then filtering accuracy is improved, but coding complexity increases
Solution Approach 1:
The patent applies preliminary action by performing motion compensation first and then applying filter information in the post-processing stage. This sequence allows the filter information to be applied to already compensated pixels, maintaining filtering accuracy while simplifying the coding process since the same filter information can be reused across layers without requiring complex layer-specific coding.
Solution Approach 2:
The patent uses motion-compensated pixels as an intermediary between the reference layer and the enhancement layer. The filter information is applied to the reference layer pixels, which are then motion compensated to generate the enhancement layer pixels. This intermediary approach maintains filtering accuracy while simplifying the overall coding structure.
3Loss of information
If filter information is reused across layers without transmission, then transmitted information amount is reduced, but filtering performance degrades
Solution Approach 1:
The patent implements feedback by using the filter information from the reference layer to improve the quality of motion-compensated pixels in the enhancement layer. The filtered reference pixels serve as a feedback source that enhances the prediction accuracy, ensuring that filtering performance is maintained or improved while reusing the same transmitted filter information across layers.
Data Source
AI summary
Provided is a scalable video-decoding method based on multiple layers. The scalable video-decoding method according to the present invention comprises: a step of predicting first filter information of a video to be filtered using the information contained in an object layer and/or information contained in another layer, and generating second filter information in accordance with the prediction; and a step of filtering the video to be filtered using the second filter information. According to the present invention, the amount of information being transmitted is reduced, and video compression performance is improved.