Spatially Extended Reference Pictures for Video Coding Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently utilizing spatial extrapolation as a predictor to enhance compression efficiency and improve picture quality, particularly in scenarios involving extended field-of-view (FOV) pictures.

Innovation Solution

The method involves forming an extended FOV picture, deriving a prediction signal using this picture, and applying it to encode or decode parts of a current source picture, leveraging global motion to spatially extend core pictures and derive prediction signals for improved encoding and decoding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If spatial extrapolation is used to extend reference pictures for prediction, then compression efficiency and picture quality are improved, but the complexity of the encoding and decoding process increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoding and decoding process complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing spatial extrapolation on reference pictures before they are used for prediction. The encoder pre-extends the reference pictures using extrapolation filters to create extended reference pictures, which are then stored and reused for predicting multiple current pictures. This preliminary preparation reduces the need for repeated complex operations during actual encoding, thereby improving compression efficiency while managing complexity.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by creating extended reference pictures that are generated once and then copied/reused for predicting multiple current pictures. The extended reference pictures serve as reusable prediction sources, eliminating the need to perform spatial extrapolation repeatedly for each current picture, thus improving productivity while controlling device complexity.

Inventive Principle:
Principle #26Copying

2Measurement precision

If spatial extrapolation is applied to extend core pictures, then the prediction accuracy for current pictures is improved, but the computational resources and processing time required increase

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

Solution Approach 1:

The patent performs spatial extrapolation as a preliminary action to extend core reference pictures before they are needed for prediction. By pre-computing the extended reference pictures and storing them, the system avoids repeating the computationally intensive extrapolation process for each current picture, thereby maintaining high prediction accuracy while reducing processing time and computational resource consumption during actual encoding/decoding operations.

Inventive Principle:
Principle #10Preliminary action

3Manufacturing precision

If extended FOV pictures are formed and used for prediction, then the quality of reconstructed pictures is improved, but the memory requirements and data processing load increase

Engineering Contradiction:
Improvepicture qualityVSAvoidmemory requirements
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies local quality by performing spatial extrapolation selectively on specific regions of reference pictures that are most beneficial for prediction. Rather than uniformly extending entire pictures, the method focuses computational resources on extending areas that will provide the most improvement in prediction accuracy, thereby achieving high picture quality while minimizing the increase in memory requirements and data processing load.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4687343A1Spatial extrapolation as predictor in video coding
Publication Date: 2026.02.04 NOKIA TECHNOLOGIES OY
  • EP4687343A1 patent drawingFigure 1
  • EP4687343A1 patent drawingFigure 2
  • EP4687343A1 patent drawingFigure 3

AI summary

A method comprising: forming an extended field-of-view (FOV) picture; deriving a prediction signal by using the extended-FOV picture; and using the prediction signal to encode at least a part of a current source picture to a current coded picture.