Hybrid Boundary Padding for Video Block Artifacts

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards face issues with boundary padding, leading to absent, unusable, or poor-quality motion information, resulting in block artifacts when handling areas outside the reference frame, especially in omnidirectional content.

Innovation Solution

The proposed solution involves a hybrid approach that combines inter-prediction-based boundary padding with global padding using overall camera motion and intra-prediction-based methods, along with overlapping block motion compensation and constrained intra-prediction to mitigate block artifacts and improve sample quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If inter-prediction-based boundary padding is used to extend the reference frame, then motion information can be obtained for areas outside the reference frame, but block artifacts occur due to inconsistent motion information and poor sample quality

Engineering Contradiction:
Improvemotion information availabilityVSAvoidblock artifacts
Core Design Contradiction:
Loss of informationVSObject-generated harmful factors

Solution Approach 1:

The patent applies different boundary padding methods to different regions of the reference frame based on their specific characteristics. For regions near the reference frame boundary, inter-prediction-based padding is used to maintain motion information. For regions where motion information is unavailable or unreliable, intra-prediction-based padding is applied to ensure consistent sample quality. This localized approach prevents block artifacts while maintaining the benefits of motion information where available.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent introduces an intermediary mechanism that combines both inter-prediction-based and intra-prediction-based boundary padding methods. This hybrid approach acts as a mediator between the two padding techniques, allowing the system to leverage the strengths of both: motion information from inter-prediction and sample quality consistency from intra-prediction, thereby reducing block artifacts.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If the reference frame area is extended to include areas outside the original frame boundaries, then motion vectors can point to valid locations, but the quality of padding samples deteriorates

Engineering Contradiction:
Improvemotion vector flexibilityVSAvoidsample quality
Core Design Contradiction:
Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent extends the reference frame area to accommodate flexible motion vectors while maintaining sample quality through localized padding strategies. Different regions of the extended frame receive different types of padding: inter-prediction-based padding in regions where motion information is available, and intra-prediction-based padding in regions where sample quality is critical. This ensures both motion vector flexibility and sample quality precision.

Inventive Principle:
Principle #3Local quality

3Device complexity

If a single boundary padding method is used for the entire reference frame, then processing is simplified, but block artifacts increase due to inconsistent padding quality

Engineering Contradiction:
Improvepadding processing complexityVSAvoidblock artifacts
Core Design Contradiction:
Device complexityVSObject-generated harmful factors

Solution Approach 1:

The patent segments the reference frame into different regions based on their proximity to boundaries and the availability of motion information. This segmentation allows the application of different padding methods to different segments: inter-prediction-based padding for regions with motion information and intra-prediction-based padding for regions without. This segmented approach reduces block artifacts while maintaining manageable processing complexity through systematic region-based processing.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3818713B1Apparatus for block-based predictive video coding
Publication Date: 2024.01.10 FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV
  • EP3818713B1 patent drawingFigure 1
  • EP3818713B1 patent drawingFigure 2
  • EP3818713B1 patent drawingFigure 3

AI summary

Apparatus for block-based predictive video decoding a video from a data stream using motion compensated prediction for inter-predicted blocks, configured to perform a once- per-picture boundary padding of a picture boundary extension area of a reference picture to obtain a pre-filling of the picture boundary extension area, check whether the pre-filling of the picture boundary extension area at a portion of a motion compensated footprint of an inter-predicted block of a current picture, which extends beyond a boundary of the reference picture, fulfills a predetermined criterion, if the portion fulfills the predetermined criterion, use the pre-filling for determining a predictor for the inter-predicted block, and if the portion does not fulfill the predetermined criterion, perform an alternative boundary padding to fill the portion to obtain a specific filling of the portion and use the specific filling for determining the picture.