Implicit Transform Splitting for Video Decoding Bandwidth Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In video coding, transform units (TUs) with only zero coefficients unnecessarily increase the bandwidth of the coded video bitstream, as inverse transforms are performed on these units despite them not containing necessary information for reconstruction.
Innovation Solution
A method is introduced where a video decoder acquires transform block signaling information to determine if TUs contain non-zero coefficients, and only performs inverse transforms on those units, using specified coding tools, thereby reducing unnecessary processing and signaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inverse transform is performed on all transform units (TUs) including those with only zero coefficients, then reconstruction accuracy is maintained, but bandwidth of the coded video bitstream increases unnecessarily
Solution Approach 1:
The patent extracts and identifies TUs containing non-zero coefficients using signaling information, then separates the processing path: only these identified TUs undergo inverse transform while zero-coefficient TUs are skipped. This extraction principle allows the system to eliminate unnecessary processing of zero-coefficient blocks while maintaining reconstruction accuracy for blocks that actually contain data.
Solution Approach 2:
The patent applies different processing quality to different TUs based on their local characteristics: TUs with non-zero coefficients receive full inverse transform processing to ensure reconstruction accuracy, while zero-coefficient TUs receive no processing. This local quality differentiation optimizes bandwidth by avoiding uniform processing of all TUs when local analysis shows some are unnecessary.
2Reliability
If inverse transform is performed on all TUs, then complete reconstruction is ensured, but processing complexity and time increase
Solution Approach 1:
The patent extracts the subset of TUs that require processing by checking signaling information for non-zero coefficient indicators. By taking out only the necessary TUs from the complete set and processing only those, the system maintains reconstruction completeness for relevant blocks while eliminating redundant processing operations on zero-coefficient blocks, thereby improving decoding speed.
Solution Approach 2:
Instead of performing inverse transform on all TUs (excessive action), the patent applies partial action by processing only the subset of TUs that contain non-zero coefficients. This partial processing approach is sufficient to achieve complete reconstruction of meaningful data while reducing overall processing complexity and time.
3Loss of information
If signaling information is transmitted for all TUs, then decoder can identify non-zero coefficient blocks, but bitstream size increases
Solution Approach 1:
The patent extracts only the essential signaling information needed to identify TUs with non-zero coefficients, rather than transmitting complete data for all TUs. This selective extraction of signaling information provides the decoder with sufficient knowledge to locate and process only necessary blocks, minimizing bitstream size while preventing information loss about where actual data resides.
Data Source
AI summary
A method of video decoding for a decoder includes acquiring transform block signaling information from a coded video bitstream. The method further includes determining whether the transform block signaling information indicates that at least one of a plurality of transform block units (TUs) contains non-zero coefficients, the plurality of TUs being partitioned from a coding block unit (CU). The method further includes, in response to the determination that at least one of the plurality of TUs contains non-zero coefficients, (i) determining a coding tool specified for one of the at least one of the plurality of TUs, and (ii) performing, for each of the at least one of the plurality of TUs that contains non-zero coefficients, an inverse transform in accordance with the coding tool specified for the one of the at least one of the plurality of TUs.