NAL Unit Header Bit Efficiency via Picture Type Inference

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional video compression technologies inefficiently use bits in the NAL unit header due to fixed assignment of nal_ref_flag and nal_unit_type, limiting the transmission of information and requiring explicit inclusion of nal_ref_flag, which is not necessary for all NAL unit types.

Innovation Solution

Infer the value of nal_ref_flag from the NAL unit type, eliminating the need for explicit inclusion in the header by determining picture types that indicate reference usage, allowing for more efficient bit allocation and extension of NAL unit types from 6 to 7 bits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If nal_ref_flag is explicitly included in the NAL unit header, then the reference information is clearly indicated, but the header size increases and bit efficiency decreases

Engineering Contradiction:
Improveinformation transmission efficiencyVSAvoidheader structure complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The invention extracts the nal_ref_flag from the NAL unit header structure and determines its value through inference based on nal_unit_type and picture type, rather than including it as an explicit field. This removes the redundant element while preserving the necessary information through logical deduction.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The NAL unit type and picture type fields serve a dual purpose: they identify the unit type and simultaneously enable inference of the nal_ref_flag value. The system uses existing fields to provide additional information without requiring extra space, making the existing structure serve multiple functions.

Inventive Principle:
Principle #25Self-service

2Adaptability or versatility

If NAL unit types are extended from 6 to 7 bits, then more NAL unit types can be defined, but the header size increases

Engineering Contradiction:
ImproveNAL unit type diversityVSAvoidheader bit length
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The invention changes the parameter of nal_unit_type from 6 bits to 7 bits, expanding the possible values from 64 to 128. This allows for greater diversity in NAL unit types while the corresponding inference mechanism for nal_ref_flag ensures that the overall header efficiency is maintained by eliminating the separate flag field.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If fixed bit allocation is used for nal_ref_flag and nal_unit_type, then the header structure is simple, but the information transmission capability is limited

Engineering Contradiction:
Improveheader structure simplicityVSAvoidinformation transmission capability
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The invention makes the picture type field serve multiple functions: it identifies the picture type and simultaneously determines the nal_ref_flag value through inference rules. This multi-functionality increases information transmission capability without adding header complexity, as the same field provides both classification and reference information.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4207765A1Video predictive encoding device, video predictive encoding method, video predictive encoding program, video predictive decoding device, video predictive decoding method, and video predictive decoding program
Publication Date: 2023.07.05 NTT DOCOMO INC
  • EP4207765A1 patent drawingFigure 1
  • EP4207765A1 patent drawingFigure 2
  • EP4207765A1 patent drawingFigure 3

AI summary

A NAL unit header in the conventional method is inefficient, even in cases where a value of nal ref flag is uniquely determined according to a value of nal_unit_type, since respective bits are assigned to nal_ref_flag and nal unit type. A solution to this problem is a video predictive encoding device provided with an input device to input pictures forming a video sequence, and an encoding unit to encode the pictures by either intra prediction or inter prediction to generate compressed picture data, and to packetize the compressed image data along with packet header information. The packet header information contains a picture type, and wherein the encoding unit determines the picture type so as to uniquely indicate whether encoded picture data is used for reference in decoding of another picture. Reference should be made to Fig. 1.