Video Coding Latency Reduction via Bounding Block Pixel Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-Efficiency Video Coding (HEVC) faces challenges in reducing latency when computing matching costs for reordering prediction candidates, particularly due to the need for frequent DRAM access and retrieval of recently reconstructed pixels, which increases processing time.
Innovation Solution
The method involves identifying major reference pictures, defining bounding blocks for these pictures to minimize DRAM access, and computing matching costs using only available pixels within these blocks, with unavailable pixels replaced by padding values or scaled from available pixels, allowing for efficient reordering of prediction candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If matching costs are computed for all prediction candidates using all reference pixels, then coding precision is improved, but processing time and latency increase due to frequent DRAM access
Solution Approach 1:
The patent segments the reference pixels into two categories: available pixels (within the bounding block of the current block in the reference picture) and unavailable pixels (outside the bounding block). Only available pixels are used for computing matching costs, while unavailable pixels are excluded. This segmentation reduces the number of DRAM accesses required while maintaining sufficient coding accuracy through alternative mechanisms such as candidate reordering and weighted cost computation.
Solution Approach 2:
The patent performs preliminary identification of the bounding block and pre-determination of which pixels are available versus unavailable before computing matching costs. This preliminary action allows the encoder to avoid unnecessary DRAM accesses for unavailable pixels, reducing latency. The bounding block is defined based on the current block's position and size, and this information is prepared in advance to guide the matching cost computation efficiently.
2Productivity
If all reference pixels are retrieved from DRAM for matching cost computation, then coding efficiency is improved, but DRAM access frequency and power consumption increase
Solution Approach 1:
The patent extracts only the necessary subset of reference pixels (available pixels within the bounding block) that are required for matching cost computation, excluding unavailable pixels that would require additional DRAM access. This extraction principle reduces energy consumption by minimizing DRAM access operations while maintaining sufficient coding efficiency through the reordering of prediction candidates based on the computed matching costs.
3Measurement precision
If prediction candidates are reordered based on complete matching cost computation, then candidate selection accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent applies partial action by computing matching costs using only available pixels rather than all reference pixels. This partial computation is sufficient to achieve effective prediction candidate reordering, as the most significant matching information is captured within the bounding block. The approach avoids the excessive complexity of processing all pixels while maintaining adequate selection accuracy through the reordering mechanism and weighted cost computation.
Data Source
AI summary
For each prediction candidate of a set of one or more prediction candidates of the current block, a video coder computes a matching cost between a set of reference pixels of the prediction candidate in a reference picture and a set of neighboring pixels of a current block in a current picture. The video coder identifies a subset of the reference pictures as major reference pictures based on a distribution of the prediction candidates among the reference pictures of the current picture. A bounding block is defined for each major reference picture, the bounding block encompassing at least portions of multiple sets of reference pixels for multiple prediction candidates. The video coder assigns an index to each prediction candidate based on the computed matching cost of the set of prediction candidates. A selection of a prediction candidate is signaled by using the assigned index of the selected prediction candidate.