Video Slice Address Signaling Overhead Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like H.264/MPEG-4 AVC and HEVC face inefficiencies in managing slice start addresses, leading to significant overhead in coded picture data due to the need for lengthy slice address signaling and calculation, especially with hierarchical coding units.
Innovation Solution
A method is introduced where each slice in a picture is assigned a unique slice flag, with the first slice having a defined start position, and remaining slices having slice addresses generated to identify their start positions within the picture, reducing the need for slice address signaling at the decoder and allowing flexible slice start positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If slice addresses are generated for all slices to identify start positions, then slice positioning accuracy is improved, but coded picture data overhead increases
Solution Approach 1:
The picture is divided into multiple slices that can be independently decoded. Each slice except the first is assigned a slice address to identify its start position. The first slice implicitly starts at position (0,0) without requiring an address, thereby reducing overhead while maintaining positioning accuracy for all slices.
Solution Approach 2:
The slice address information is extracted from the general slice header structure and handled separately. The first slice's start position is extracted as the implicit origin (0,0), eliminating the need to encode and transmit its address, thereby reducing the total amount of slice address data in the coded picture representation.
2Productivity
If hierarchical coding units (LCUs) are used to improve coding efficiency, then compression performance is improved, but slice address management complexity increases
Solution Approach 1:
Different slices have different address management requirements. The first slice is treated differently from subsequent slices by using an implicit (0,0) start position. This local differentiation simplifies the overall address management system while maintaining the flexibility needed for hierarchical LCUs in different regions of the picture.
Solution Approach 2:
The first slice's start position is predetermined as (0,0) before encoding begins. This preliminary assignment eliminates the need for complex address calculation and transmission for the first slice, simplifying the address management system while allowing hierarchical LCUs to be efficiently organized from the picture origin.
3Reliability
If slice boundaries are aligned with LCU borders, then decoding independence is improved, but flexibility in slice positioning is reduced
Solution Approach 1:
The slice address system allows dynamic positioning of slice boundaries at LCU borders while maintaining the ability to adapt to different picture sizes and LCU configurations. Each slice address points to a specific LCU border position, enabling flexible slice placement that maintains decoding independence while adapting to various content requirements.
Data Source
Figure 1~4
Figure 3A~8
Figure 6~11
AI summary
A picture with multiple slices is encoded by generating (S1) a coded slice representation for each of the slices. A slice flag is set (S2) to a first value for the first slice in the picture and corresponding slice flags of the remaining slices are set (S3) to a second defined value. A respective slice address is generated (S4) for each remaining slice to enable identification of the slice start position within the picture for the slice. The slice address is a fixed length address. A coded picture representation of the picture is generated (S5) which comprises the coded slice representations, the slice addresses, and the slice flags. The slice flags enable differentiation between slices for which slice addresses are required and the slice per picture for which no slice address is needed to identify its slice start position.