External Boundary Smoothing for Parallel Decoded Image Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression standards face challenges in smoothing boundaries between image regions processed by different encoding cores, leading to visual artifacts and asymmetrical processing that induces drift, which is costly in terms of computation time and requires modifications to existing coding algorithms.
Innovation Solution
A method for processing decoded image regions using metadata to implement boundary smoothing outside the decoding loop, ensuring symmetry between encoder and decoder processing without impacting future decoding, and avoiding visible boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If boundary smoothing is applied inside the decoding loop to improve boundary visibility, then boundary artifacts are reduced, but encoder-decoder symmetry is broken and drift is induced
Solution Approach 1:
The boundary smoothing operation is extracted from the decoding loop and applied outside it. The patent applies the smoothing filter to the decoded picture buffer after the standard decoding loop completes, using a separate processing stage that does not feed back into the decoding loop. This extraction eliminates the asymmetry and drift problems while still achieving boundary smoothing.
Solution Approach 2:
The patent introduces an intermediary processing stage between the decoding loop and the final output. The boundary smoothing is applied as a post-processing step that takes the decoded picture as input and produces the final smoothed output, without being part of the feedback loop. This intermediary position preserves encoder-decoder symmetry while achieving the desired smoothing effect.
2Reliability
If boundary smoothing is applied at the encoder to maintain symmetry, then encoder-decoder symmetry is preserved, but additional computation time and algorithm modifications are required
Solution Approach 1:
Instead of applying boundary smoothing at the encoder side to maintain symmetry, the patent inverts the approach by applying the smoothing at the decoder side outside the decoding loop. This inversion allows the use of already-decoded data without requiring encoder modifications or additional encoding computation time.
Solution Approach 2:
The patent performs boundary smoothing as a preliminary post-processing step immediately after decoding, using the decoded picture buffer data that is already available. This preliminary action utilizes existing decoded data without requiring additional encoding computations or algorithm modifications at the encoder.
3Productivity
If parallel processing with multiple encoding cores is used to improve productivity, then encoding speed is increased, but boundary smoothing becomes complex and requires data sharing between cores
Solution Approach 1:
The patent segments the boundary smoothing operation into independent per-block processing that can be performed after parallel encoding. Each block at the boundary is smoothed independently using its own decoded data, allowing the smoothing to be distributed and processed without requiring complex inter-core data sharing mechanisms.
Solution Approach 2:
The boundary smoothing process is designed to be self-sufficient, using only the decoded picture buffer data that is already available after parallel encoding. Each boundary block retrieves and processes its own data without requiring data sharing or synchronization between encoding cores, simplifying the implementation while maintaining parallel processing benefits.
Data Source
AI summary
A method for processing at least one decoded image region. At the exit of a decoding loop that has decoded at least one current image region, the decoded current image region is processed with the aid of a boundary-smoothing module that uses metadata relating at least to a boundary between the current image region and a neighboring image region.