Single-Pass Deblocking Filter for WMV Video Decoders
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current deblocking filters in video decoders are computationally intensive and require multiple passes, leading to high memory access times and power consumption, while existing algorithms are complex and difficult to implement in real-time MPEG-4 decoders.
Innovation Solution
A deblocking module that performs filtering in one pass on small data blocks, reducing external memory traffic and using a single-pass scheme for both luma and chroma on 8×8 or 4-wide boundaries, with a hardware implementation that processes one 16×16 macroblock at a time, optimizing computation and memory access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple passes are used for deblocking filtering, then filtering quality is improved, but memory access time and power consumption increase
Solution Approach 1:
The patent divides the image into small processing units (e.g., 8×8 pixel blocks) and processes each segment independently in a single pass. This segmentation allows the filter to work on localized regions with minimal memory access, avoiding the need to repeatedly access external memory for entire image passes while still achieving effective deblocking filtering on each segment.
Solution Approach 2:
The patent changes the processing dimension by operating on small spatial blocks simultaneously rather than processing the entire image sequentially in multiple passes. This dimensional approach to processing allows all necessary filtering operations to be completed in one pass through the image data, reducing total memory access time.
2Manufacturing precision
If multiple passes are used for deblocking filtering, then filtering quality is improved, but power consumption increases
Solution Approach 1:
By segmenting the image into small processing blocks and filtering each in a single pass, the patent eliminates the repeated activation of the processing unit and external memory interface that would occur in multiple-pass approaches. This single-pass segmented processing significantly reduces overall power consumption while maintaining filtering effectiveness on each block.
Solution Approach 2:
The patent implements continuous single-pass filtering where processing occurs continuously as data flows through the system, rather than repeatedly accessing memory between filtering operations. This continuous action eliminates idle power consumption periods between passes while maintaining filtering quality through the segmented block processing approach.
3Manufacturing precision
If complex deblocking algorithms are used, then block artifact removal is improved, but implementation complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the image into small 8×8 pixel blocks and applying a simplified filtering operation to each block independently. This segmentation allows complex-looking filtering to be achieved through simple, repetitive operations on small units, reducing the complexity of individual processing elements while maintaining overall filtering effectiveness.
Solution Approach 2:
The patent applies filtering to all 8×8 pixel blocks uniformly, including those that may not strictly require deblocking. This partial or excessive application of the filter ensures that all potential block artifacts are addressed with a simple, consistent operation, avoiding the need for complex edge detection and adaptive filtering logic.
4Loss of time
If small data blocks are processed in one pass, then memory access time is reduced, but processing speed may be affected
Solution Approach 1:
The patent segments the image into small 8×8 pixel blocks that can be processed independently in a single pass through the memory. This segmentation minimizes the amount of data that needs to be held in memory during processing, reducing memory access time while the systematic processing of all segments maintains overall processing throughput.
Solution Approach 2:
The patent merges the filtering operation with the single-pass data flow through the system. By combining the deblocking filter with the existing data processing pipeline and processing all blocks in one pass, the patent avoids the overhead of multiple separate processing stages while maintaining processing speed through efficient data utilization.
Data Source
AI summary
Deblock filtering for Microsoft WMV video decoders partitions the computation so that the deblock filtering operations can be performed on horizontal or vertical stripes or in one pass on oversized macroblocks.


