Multi-Layer Video Bitstream Reference Layer Signaling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding technologies face inefficiencies in specifying reference layers for multi-layer bitstreams, leading to suboptimal coding efficiency and user experience.

Innovation Solution

Implementing a method to derive and update a reference layer flag based on a direct reference layer flag, ensuring any reference layers are specified for a layer, thereby improving coding efficiency by quickly identifying reference layers during the coding process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If reference layers are specified using existing methods in multi-layer bitstreams, then the bitstream can be decoded, but coding efficiency is suboptimal due to redundant signaling and difficulty in quickly identifying reference layers

Engineering Contradiction:

Improvecoding efficiencyVSAvoidtime to identify reference layers

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by introducing a new syntax element `direct_ref_layer_idx` in the VPS (Video Parameter Set) that pre-specifies direct reference layer relationships before decoding begins. This allows the decoder to quickly identify reference layers without extensive searching during the decoding process, thereby improving coding efficiency and reducing the time loss in identifying reference layers.

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If more reference layer information is signaled in the bitstream, then reference layers can be accurately identified, but the bitstream size increases leading to redundant signaling

Engineering Contradiction:

Improvereference layer identification accuracyVSAvoidbitstream size

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential reference layer information needed for decoding by introducing a targeted syntax element `direct_ref_layer_idx` that specifically identifies direct reference layers. This extraction approach provides accurate reference layer identification without signaling all possible reference layer relationships, thereby reducing redundant signaling and keeping the bitstream size manageable.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter representation by using an indexed approach (`direct_ref_layer_idx`) rather than signaling complete reference layer lists. This parameter change enables precise reference layer identification with fewer bits, reducing the quantity of data in the bitstream while maintaining identification accuracy.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12563236B2Avoidance of redundant signaling in multi-layer video bitstreams

Publication Date: 2026.02.24 HUAWEI TECH CO LTD

AI summary

A method of decoding includes receiving a bitstream having a plurality of layers and a video parameter set (VPS), where the VPS includes a direct reference layer flag, where the direct reference layer flag equal to a first value specifies that a second layer is not a direct reference layer for a first layer, and where the direct reference layer flag equal to a second value specifies that the second layer is the direct reference layer for the first layer; deriving a reference layer flag by: initially setting the reference layer flag to the first value; and updating the reference layer flag to the second value when the direct reference layer flag has the second value; and decoding a picture from the first layer in accordance with the reference layer flag to obtain a decoded picture.