Image Decoding With Parallel Intra-Subpartition Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Intra Sub-Partitions (ISP) processing requires sequential execution of sub-blocks, leading to a bottleneck in processing speed when implemented in hardware.
Innovation Solution
An image decoding device that generates prediction signals by intra prediction, allowing parallel processing of sub-blocks within a block, with reference to adjacent decoded blocks, and optimizing block division and intra prediction mode decoding based on block shape.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intra prediction of sub-blocks is performed sequentially by referring to decoded sub-blocks, then prediction accuracy is improved, but processing speed deteriorates
Solution Approach 1:
The block is divided into multiple sub-blocks, and prediction signals are generated independently for each sub-block by referring to pixels from adjacent decoded blocks rather than sequentially from previously processed sub-blocks. This segmentation enables parallel processing of multiple sub-blocks simultaneously, resolving the contradiction between prediction accuracy and processing speed.
2Measurement precision
If intra sub-partitions are applied to divide blocks, then prediction precision is improved, but device complexity increases
Solution Approach 1:
The block is divided into multiple sub-blocks with different shapes (e.g., rectangular, triangular, L-shaped) based on the characteristics of the original block. Each sub-block is processed independently with simplified prediction operations, which maintains prediction precision while reducing overall processing complexity through parallel execution.
Solution Approach 2:
Different sub-blocks within the same block can have different shapes and prediction modes selected according to local image characteristics. This allows the system to adapt to local variations in the image data, improving prediction precision without uniformly increasing complexity across the entire block.
3Device complexity
If sequential processing of sub-blocks is used, then processing logic is simplified, but productivity decreases
Solution Approach 1:
The processing is segmented into independent sub-block operations that can be executed in parallel. Each sub-block processing unit operates independently using the same simplified logic, but the overall productivity increases because multiple sub-blocks are processed simultaneously rather than sequentially.
Solution Approach 2:
The same prediction processing logic is copied and applied to multiple sub-blocks independently. Instead of having complex sequential dependencies, the system uses identical processing templates for each sub-block, which can be instantiated and executed in parallel, thereby improving productivity without complicating the base processing logic.
Data Source
AI summary
An image decoding device includes: a prediction signal generation unit that generates a prediction signal by intra prediction, wherein in a block to which intra sub-partitions are applied, the prediction signal generation unit refers to a pixel of a decoded block adjacent to the block when generating prediction signals of all sub-blocks in the block.