Multi-layer Video Decoding Reference Picture Management

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing multi-layer video decoding methods face inefficiencies in managing the decoded picture buffer and transmitting reference picture information, particularly in scalable video encoding (SVC) and multi-view video coding (MVC), which affects the performance of interlayer prediction and picture quality.

Innovation Solution

A method for decoding multi-layer video that involves marking and updating the reference picture status of pictures in the decoded picture buffer based on interlayer reference picture set information, allowing for efficient interlayer prediction and buffer management by marking pictures as short-term or long-term reference pictures based on their usage and picture order count.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If reference picture information is transmitted for every picture in multi-layer video, then prediction accuracy is improved, but buffer management complexity and transmission overhead increase

Engineering Contradiction:
Improveprediction accuracyVSAvoidbuffer management complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and transmits only the necessary reference picture information (interlayer RPS information) rather than all reference picture data. This selective extraction reduces transmission overhead while maintaining prediction accuracy by focusing only on the most relevant reference pictures for interlayer prediction.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different reference picture management strategies to different layers and picture types. Short-term reference pictures are managed differently from long-term reference pictures, and interlayer reference information is provided selectively based on the specific needs of each enhancement layer, optimizing both accuracy and complexity.

Inventive Principle:
Principle #3Local quality

2Productivity

If more reference pictures are stored in the decoded picture buffer, then interlayer prediction performance is improved, but buffer memory requirements and management overhead increase

Engineering Contradiction:
Improveinterlayer prediction performanceVSAvoidbuffer memory requirements
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent changes the temporal validity parameters of reference pictures by distinguishing between short-term and long-term reference pictures. This allows the buffer to maintain high prediction performance by keeping essential reference pictures longer while automatically removing less important ones, optimizing the balance between performance and memory usage.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary classification of reference pictures into short-term and long-term categories before they are stored in the buffer. This preliminary action ensures that only necessary reference pictures are retained, preventing buffer overflow while maintaining prediction performance through proactive memory management.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If reference picture status is frequently updated, then decoding accuracy is improved, but processing time and computational overhead increase

Engineering Contradiction:
Improvedecoding accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements periodic updates of reference picture status based on picture order count (POC) values and layer information. Rather than continuous updates, the system periodically refreshes reference picture lists at strategically determined points, maintaining decoding accuracy while reducing processing overhead through rhythmic, predictable update cycles.

Inventive Principle:
Principle #19Periodic action

Data Source

PatentUS9967574B2Method and apparatus for decoding multi-layer video, and method and apparatus for encoding multi-layer video
Publication Date: 2018.05.08 SAMSUNG ELECTRONICS CO LTD

AI summary

A method of decoding multi-layer video includes decoding a first layer picture, storing the decoded first layer picture in a decoded picture buffer (DPB), and marking the first layer picture as a short-term reference picture; obtaining interlayer reference picture set (RPS) information regarding a second layer picture having a same first picture order count (POC) as the first layer picture; marking the first layer picture, which is marked as the short-term reference picture, as a long-term reference picture, based on the interlayer RPS information; performing interlayer prediction with respect to the second layer picture with reference to the first layer picture; decoding the second layer picture based on a result of performing the interlayer prediction; and, after the second layer picture is decoded, marking the first layer picture, which is marked as the long-term reference picture, as a short-term reference picture.