In-Loop Filter Buffer Reduction via Virtual Boundary Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In video coding systems, the use of line buffers for in-loop processing of Sample Adaptive Offset (SAO) and Adaptive Loop Filter (ALF) requires significant memory and increases power consumption and chip cost, especially in hardware-based implementations, due to the need for large buffers to store processed pixels across LCU boundaries.
Innovation Solution
The implementation of a virtual boundary processing method that reduces or eliminates the need for pixels from the other side of the virtual boundary during SAO and ALF processing, using replacement pixels, padding, or averaging to generate filtered outputs, thereby minimizing line buffer requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If large line buffers are used to store processed pixels across LCU boundaries for SAO and ALF processing, then processing completeness and picture quality are improved, but memory requirements, power consumption, and chip cost increase significantly
Solution Approach 1:
The patent divides the processing into two distinct phases: an odd-line processing phase that processes lines 0, 2, 4, etc. without requiring pixels from the other side of the virtual boundary, and an even-line processing phase that processes lines 1, 3, 5, etc. The segmentation allows each phase to operate with reduced buffer requirements, as the odd-line phase does not need to store pixels that would otherwise require buffering for complete boundary processing.
Solution Approach 2:
The patent performs preliminary processing of odd lines before even lines, and completes certain processing operations before moving to the next phase. By processing odd lines first and preparing certain data structures in advance, the system reduces the peak memory requirements needed during even-line processing, as some data is already prepared or can be generated on-demand rather than requiring full buffering.
2Reliability
If large line buffers are used to store processed pixels across LCU boundaries, then in-loop filter performance is improved, but power consumption increases
Solution Approach 1:
By segmenting the processing into odd-line and even-line phases, the system reduces the peak buffer size required, which directly reduces the power consumption of memory access operations. The segmented approach allows the buffer to be smaller and operate at lower capacity, reducing dynamic power consumption proportional to the reduced memory access volume and smaller buffer size.
3Reliability
If large line buffers are used for SAO and ALF processing, then picture quality is improved, but chip cost increases
Solution Approach 1:
The patent segments the line buffer into smaller functional units that process odd and even lines separately. This segmentation reduces the total buffer size required from 8.3 lines for luma and 6.2 lines for chroma to significantly smaller amounts, reducing the SRAM area required on the chip. The smaller buffer area directly reduces manufacturing cost while maintaining picture quality through the phased processing approach.
Data Source
Figure 1A
Figure 1B
Figure 2A~2D
AI summary
A method and apparatus for in-loop processing of reconstructed video are disclosed. The method and apparatus configure the in-loop processing so that the processing requires no pixel or reduced pixels from other side of a virtual boundary. When the in-loop processing of the to- be- processed pixel requires a pixel from the other side of the virtual boundary, the pixel from the other side of the virtual boundary is replaced by a replacement pixel. The in-loop processing can also be configured to skip the pixel when the processing requires a pixel from other side of the virtual boundary. The in-loop processing can also be configured to change ALF filter shape or filter size when the in-loop processing requires a pixel from other side of the virtual boundary. A filtered output can be combined linearly or nonlinearly with the to-be- processed pixel to generate a final filter output.