Scalable Video Coding Error Concealment via Base Layer Upsampling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional error concealment methods in Scalable Video Coding (SVC) render poor picture quality when packet loss occurs in enhancement layers, especially during scene changes, due to inadequate protection and propagation of errors across GOPs.

Innovation Solution

The method involves replacing lost pictures in the target layer with up-sampled pictures from a lower layer, such as the base layer, during both single-loop and multi-loop decoding, to minimize picture degradation by ignoring or using up-sampled base layer pictures until an I-picture key frame is received, thereby stopping error propagation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional error concealment methods are used to replace lost pictures, then packet loss in enhancement layers can be handled, but picture quality deteriorates significantly during scene changes

Engineering Contradiction:
Improveerror concealment capabilityVSAvoidpicture quality
Core Design Contradiction:
ReliabilityVSManufacturing precision

Solution Approach 1:

The video stream is segmented into base layer and enhancement layers, allowing independent decoding of the base layer and selective use of enhancement layer pictures. When packet loss occurs in enhancement layers, the base layer pictures can be independently decoded and used for error concealment, maintaining picture quality during scene changes while still benefiting from enhancement layer protection when available.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different protection strategies are applied to different layers: the base layer receives greater protection with lower packet loss probability, while enhancement layers have lower protection with higher packet loss probability. This local differentiation allows the system to maintain picture quality by using base layer pictures for error concealment in enhancement layers during scene changes.

Inventive Principle:
Principle #3Local quality

2Quantity of substance

If enhancement layers are transmitted with lower protection due to bandwidth limitations, then bandwidth efficiency is improved, but packet loss probability in enhancement layers increases

Engineering Contradiction:
Improvebandwidth efficiencyVSAvoidpacket loss probability
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The base layer is transmitted with excessive protection (higher protection than strictly necessary) to ensure reliable decoding even under poor channel conditions. The enhancement layers are transmitted with partial protection (lower protection) to maximize bandwidth efficiency. This partial/excessive action strategy allows the system to maintain acceptable packet loss rates in enhancement layers while preserving the ability to perform error concealment using the well-protected base layer.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If lost pictures are replaced by duplicating previous frames, then error concealment is achieved, but picture quality deteriorates when differences between neighboring frames are large

Engineering Contradiction:
Improveerror concealment capabilityVSAvoidpicture quality during scene changes
Core Design Contradiction:
ReliabilityVSManufacturing precision

Solution Approach 1:

The base layer acts as an intermediary for error concealment in enhancement layers. Instead of directly duplicating previous enhancement layer frames (which causes poor quality during scene changes), the system uses base layer pictures as intermediaries to reconstruct lost enhancement layer pictures, thereby maintaining picture quality while still achieving error concealment.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS8798145B2Methods for error concealment due to enhancement layer packet loss in scalable video coding (SVC) decoding
Publication Date: 2014.08.05 INTERDIGITAL MADISON PATENT HLDG
  • US8798145B2 patent drawing
  • US8798145B2 patent drawing
  • US8798145B2 patent drawing

AI summary

Methods for concealing Scalable Video Coding (SVC) decoding errors. Lost pictures in the target layer are replaced with up-sampled pictures from a lower layer, such as the base layer. In multiloop decoding applications, lost pictures in the target layer are replaced with corresponding up-sampled pictures from the lower layer. Subsequent target layer pictures in the same Group of Pictures (GOP) are either replaced with corresponding upsampled pictures from the lower layer, or they are decoded with up-sampled pictures from the lower layer used as reference pictures. For single-loop decoding, the bit stream of each layer is buffered, for each GOP, and inspected for errors. If an error is detected in the target layer, the pictures of the highest level layer without an error are decoded and up-sampled and used as the output pictures for the current GOP.