Image Slice Partitioning for Wavefront Parallel Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding and decoding methods face inefficiencies when using both parallel processing and dependent slices, leading to increased complexity and potential coding losses due to dependencies between slices.
Innovation Solution
The proposed method restricts the partitioning of image rows such that a second row following a normal slice starting at a position other than the beginning does not start with a dependent slice, allowing for improved efficiency by reducing dependencies and enabling parallel processing without coding losses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If dependent slices are used to improve compression efficiency, then coding efficiency is improved, but processing complexity increases and parallel processing capability deteriorates
Solution Approach 1:
The image is divided into multiple slices, where each slice is further segmented into coding units arranged in rows. This segmentation allows independent processing of rows while maintaining slice-level dependencies, thereby improving parallel processing capability without completely sacrificing compression efficiency.
Solution Approach 2:
The patent introduces a new dimension of parallel processing by enabling row-level independence within slices. Instead of treating slices as monolithic units, the invention creates a hierarchical structure where rows can be processed in parallel within the same slice, adding a new dimension to the processing architecture.
2Loss of substance
If dependent slices are used to reduce transmission data量, then compression ratio is improved, but transmission delays increase due to dependencies
Solution Approach 1:
By segmenting slices into independently processable rows, the patent reduces the critical path length for decoding. While slice-level dependencies remain for compression efficiency, row-level independence allows overlapping processing operations, thereby reducing transmission and processing delays.
Solution Approach 2:
The patent enables preliminary decoding of rows that do not depend on current slice information. By identifying and decoding independent rows in advance, the system reduces overall transmission delays while maintaining the compression benefits of dependent slices.
3Speed
If parallel processing is implemented to improve decoding speed, then processing speed is improved, but coding complexity increases
Solution Approach 1:
The patent segments the decoding process into row-level operations that can be executed in parallel. Each row contains sufficient information for independent decoding decisions, allowing multiple rows to be processed simultaneously without requiring complex inter-row coordination logic.
Solution Approach 2:
The patent applies different processing strategies to different rows within slices. Rows that can be decoded independently receive simplified processing, while rows requiring slice-level context receive more comprehensive processing. This local differentiation reduces overall coding complexity while maintaining parallel processing capabilities.
Data Source
AI summary
An image includes a plurality of rows of largest coding units (LCUs). Encoding the image includes determining whether or not wavefront parallel processing is enabled, and partitioning the plurality of rows of LCUs to comprise a normal slice and a group of dependent slices. The normal slice includes at least one LCU, and the group of dependent slices includes at least one LCU and consists of every dependent slice that uses information from a slice header of the normal slice for encoding. Based on the determination that wavefront parallel processing is enabled, the partitioning of the plurality of rows of LCUs is restricted such that a last LCU in the group of dependent slices belongs in a same row as a first LCU of the normal slice when the first LCU of the normal slice is not a first LCU of the row.