Video Coding Picture Header Segmentation for Bitrate Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as H.264/AVC and AVS, face inefficiencies in compression and vulnerability to transmission errors due to repeated signaling of infrequently changing information in sequence and picture headers, leading to wasteful bit rate usage and potential decoding issues.
Innovation Solution
Incorporating a picture header in the picture layer and optionally in slice headers, along with parameter sets, to decouple and efficiently convey infrequently changing information, using unique start codes or NAL units to support sequence and picture parameter sets, thereby reducing redundant data transmission and enhancing error resilience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If sequence and picture headers are used to convey video coding parameters, then decoding information can be provided, but redundant data transmission occurs and bit rate is wasted
Solution Approach 1:
The patent segments header information into two distinct parts: parameter sets (containing infrequently changing information) and headers (containing frequently changing information). This segmentation allows the parameter sets to be transmitted separately and reused across multiple pictures, while headers only convey picture-specific information. The bitstream structure is divided into parameter set NAL units and picture NAL units, eliminating redundant transmission of parameter set data in every picture header.
Solution Approach 2:
The patent extracts infrequently changing parameter information from the traditional header structure and places it into separate parameter sets. These parameter sets are transmitted as independent NAL units before the picture data, allowing decoders to reuse them across multiple pictures. This extraction removes the redundant transmission of parameter set data that would otherwise occur in every picture header, reducing overall bit rate consumption.
2Reliability
If sequence and picture headers are repeated in every picture, then decoding information is available, but transmission errors have severe impact on decoding processes
Solution Approach 1:
By segmenting the bitstream into parameter set NAL units and picture NAL units, the patent isolates error-prone elements. Parameter sets are transmitted once and protected by error correction, while picture NAL units contain only picture-specific information. This segmentation prevents error propagation across multiple pictures that would occur with repeated headers, improving decoding reliability without significantly increasing structural complexity.
Solution Approach 2:
The patent implements error protection mechanisms in advance by transmitting parameter sets as separate, protected NAL units before picture data. This allows the use of more robust error correction coding on parameter sets without impacting picture decoding performance. The separate parameter set structure provides a buffer against transmission errors, cushioning the impact on overall decoding reliability.
3Productivity
If traditional header structures are used, then picture decoding information is provided, but compression efficiency is reduced due to redundant signaling
Solution Approach 1:
The patent achieves improved compression efficiency by segmenting header information into parameter sets and picture-specific headers. Parameter sets containing infrequently changing information (such as sequence parameter set and picture parameter set data) are transmitted once and reused across multiple pictures. This eliminates redundant signaling of the same parameter data in every picture header, directly improving compression efficiency by reducing the total bit rate required for header information transmission.
Data Source
AI summary
In a start code based bistream structure for use in video coding, a picture header is used simultaneously with the parameter set in the start code. The picture header is present before the slice header in the same picture. As such, it is possible to allow the presence of at least one sequence parameter set or picture parameter set before the slice header. In a NAL unit parameter set bitstream structure, the picture head NAL unit is present before the slice NAL unit. In either structure, the picture header information must remain unchanged during the decoding of the entire picture, but should change more frequently between pictures as compared to the information in the picture parameter set.