Arithmetic Entropy Coding Parallel Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding standards face inefficiencies in compression and decoding due to sequential processing of pixel probabilities, leading to delayed decoding and lower throughput, especially at high bit-rates or high-quality video processing.
Innovation Solution
Implementing a non-raster scan order for processing pixels, allowing simultaneous computation of probability model contexts within blocks rather than entire frames, and interleaving per-pixel prediction syntax with transform coefficients to reduce dependency and increase parallelism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sequential raster-order processing is used to ensure accurate probability context decoding, then decoding accuracy is maintained, but throughput and processing speed deteriorate
Solution Approach 1:
The patent segments the frame into multiple parallel processing lanes, each handling a subset of pixels. Instead of processing the entire frame sequentially in raster order, multiple lanes process different pixel groups simultaneously, each maintaining its own probability context state. This segmentation enables parallel execution while preserving the accuracy requirements of probability context decoding within each lane.
Solution Approach 2:
The patent introduces a new dimension of parallelism by processing multiple scan lines simultaneously across different lanes rather than sequentially in a single dimension. Each lane maintains independent probability context states, allowing concurrent processing of multiple rows of pixels while preserving decoding accuracy through proper state management within each lane.
2Device complexity
If per-pixel prediction syntax is separated from transform coefficients, then coding simplicity is maintained, but processing dependency increases and throughput decreases
Solution Approach 1:
The patent merges per-pixel prediction syntax and transform coefficients into a unified parallel processing structure. By interleaving these data elements in the bitstream and processing them simultaneously across multiple lanes, the patent reduces processing dependencies while maintaining coding simplicity. Each lane processes both prediction and transform data in parallel, eliminating sequential bottlenecks.
3Measurement precision
If probability context is determined by spatial neighbor pixels in raster order, then coding accuracy is improved, but decoding delay increases due to sequential dependency
Solution Approach 1:
The patent segments the probability context computation into multiple independent lanes, where each lane processes a subset of pixels with its own local neighbor references. This segmentation allows parallel computation of probability contexts across different lanes while maintaining coding accuracy through proper handling of spatial relationships within each lane's processing scope.
Solution Approach 2:
The patent performs preliminary organization of pixel data and probability context states before parallel processing begins. By pre-configuring the initial states and data arrangements for multiple lanes, the patent enables immediate parallel computation without sequential setup delays, reducing overall decoding delay while maintaining accuracy.
Data Source
AI summary
A method, system, and articles of high throughput arithmetic entropy coding for video coding uses a non-framewidth raster order or non-raster order to form spatial neighbor probability contexts for entropy coding.