Video Decoder Partitioning and Truncated Transforms for Faster Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video decoding techniques are computationally intensive, particularly for high-quality, high-bit rate video, and consume significant memory resources, leading to inefficiencies in decoding processes such as entropy decoding and inverse frequency transforms.
Innovation Solution
Implementing a multi-symbol lookup table for entropy decoding and selectively using truncated inverse frequency transforms, along with organizing blocks into partitions using restart markers to facilitate parallel decoding, reduces computational costs and optimizes resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional entropy decoding and inverse frequency transform methods are used, then decoding accuracy is maintained, but computational complexity and memory consumption increase significantly
Solution Approach 1:
The patent segments the picture into multiple partitions using restart markers, allowing independent decoding of each partition. This segmentation enables parallel processing across multiple processing cores, reducing the computational complexity burden on any single core while maintaining overall decoding accuracy through consistent reconstruction methods applied to each partition.
Solution Approach 2:
The patent performs preliminary organization of encoded data into partitions marked with restart markers during the encoding phase. This preliminary action prepares the data structure in advance to facilitate efficient parallel decoding, allowing the decoder to immediately distribute work across multiple cores without requiring complex runtime analysis or dynamic scheduling decisions.
2Manufacturing precision
If traditional decoding methods are used, then complete reconstruction is achieved, but decoding speed decreases
Solution Approach 1:
By dividing the picture into multiple independent partitions using restart markers, the patent enables simultaneous decoding of multiple partitions across different processing cores. This segmentation multiplies the effective decoding throughput while ensuring complete reconstruction through consistent inverse transforms applied to each partition, directly addressing both speed and completeness requirements.
Solution Approach 2:
Each partition is designed to be independently decodable with complete information needed for its reconstruction contained within the partition boundaries (thanks to restart markers). This self-service property allows any processing core to independently decode any partition without requiring sequential coordination, maximizing parallel processing efficiency while maintaining reconstruction completeness.
3Measurement precision
If more computing resources are allocated to decoding, then decoding quality improves, but resource consumption increases
Solution Approach 1:
The patent divides the picture into multiple smaller partitions that can be decoded and processed independently. This segmentation allows the system to process multiple partitions in parallel using the same amount of memory resources, effectively increasing the total processing capacity without requiring additional memory per core. Each partition uses a fixed, manageable amount of memory, enabling scalable deployment across systems with varying resource constraints while maintaining high decoded quality through complete reconstruction of each partition.
Data Source
AI summary
Efficient operations in image or video decoding. For example, a tool such as an image or video decoder receives and decodes encoded data for a picture in a bitstream. As part of the decoding, the tool adapts a multi-symbol lookup table to use in decoding of symbols then decodes the symbols using the multi-symbol lookup table, producing exactly correct results. The tool can also perform selectively truncated inverse frequency transforms. For a given block, the tool identifies upper horizontal and vertical frequencies among non-zero coefficients for the block and, based on the upper frequency values, selectively applies a simplified inverse frequency transform to transform coefficients for the block without hurting decoding quality. Using restart markers in the bitstream, the tool can organize multiple blocks of the picture as partitions. The tool decodes at least some of the partitions in parallel on a partition-by-partition basis using multiple processing cores.