Parallel Filtering in Video Decoding to Reduce Buffering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video decoding processes require significant memory and processing resources due to the need to buffer pre-deblocked and post-deblocked samples for adaptive loop filtering, increasing hardware complexity and processing requirements.
Innovation Solution
Configuring video encoders and decoders to perform additional filtering in parallel with deblocking filtering, combining the results to reduce memory requirements by eliminating the need to store pre-deblocked samples, using filters like guided, bilateral, or adaptive loop filters alongside deblocking filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If deblocking filtering is applied sequentially followed by adaptive loop filtering, then filtering quality is improved, but memory requirements and hardware complexity increase due to buffering pre-deblocked samples
Solution Approach 1:
The patent combines deblocking filtering and adaptive loop filtering into a single parallel processing stage. Instead of applying deblocking filter first and then ALF sequentially, both filters are applied simultaneously to the original block, and their results are combined through weighted averaging. This merging eliminates the need to buffer pre-deblocked samples, reducing memory requirements and hardware complexity while maintaining filtering quality.
Solution Approach 2:
The patent performs preliminary weighted combination of filter results during the filtering stage itself. By calculating weighted averages of deblocked and ALF results immediately after both filters produce their outputs, the patent eliminates the need for subsequent separate processing stages that would require additional buffering. This preliminary action consolidates multiple processing steps into one, reducing hardware complexity.
2Manufacturing precision
If deblocking filtering is applied sequentially followed by adaptive loop filtering, then filtering quality is improved, but buffer requirements increase due to storing pre-deblocked samples
Solution Approach 1:
The patent merges the output stages of deblocking filter and ALF by computing weighted averages simultaneously. Instead of producing separate deblocked samples that need to be buffered for later ALF processing, the patent directly computes the combined result as a weighted average of both filter outputs. This eliminates the intermediate buffer requirement for pre-deblocked samples while maintaining the quality benefits of both filtering approaches.
3Manufacturing precision
If multiple filtering stages are applied sequentially, then filtering quality is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary computation of the final filtered result by calculating weighted averages during the filtering stage itself. Instead of executing deblocking filter, buffering results, then executing ALF in sequence, the patent computes both filters in parallel and immediately combines their results through weighted averaging. This preliminary action consolidates multiple sequential processing steps into a single parallel stage, reducing processing time while maintaining filtering quality.
Data Source
AI summary
A device for decoding video data may be configured to apply a deblocking filter to a block of video data to determine a first filtered block; apply a second filter to the block of video data in parallel with the deblocking filter to determine a second filtered block, wherein the second filter comprises one of a guided filter, a bilateral filter, or an adaptive loop filter; combine the first filtered block and the second filtered block to determine a combined block; apply a third filter to the combined block to determine a third filtered block; process the third filtered block to determine a decoded version of the block of video data; and output the decoded version of the block of video data.