Video Frame Boundary Extension for Motion Vector Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding systems face inefficiencies in handling motion vectors, particularly in non-scalable, scalable, and multiview video coding, leading to sub-optimal performance in encoding and decoding processes.

Innovation Solution

The proposed method and apparatus enhance video coding by improving the handling of motion vectors through advanced prediction mechanisms, including spatial and temporal candidate sources, boundary extension techniques, and optimized motion vector prediction, which are integrated into video coding systems to improve encoding and decoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional motion estimation and compensation is used for panorama images, then the encoding process can be performed, but the performance is sub-optimal due to lack of advanced prediction mechanisms

Engineering Contradiction:
Improveencoding efficiencyVSAvoidvideo quality
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies boundary extension by padding the reference image with boundary pixels before motion estimation is performed. This preliminary action creates an extended reference image that contains pixel values at the borders, enabling more accurate motion vector calculation for blocks near the image boundaries without requiring complex post-processing or multiple estimation passes.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediate padded region with interpolated pixel values between the original image boundaries and the motion estimation process. This intermediary layer of synthetic pixel data acts as a mediator that enables smooth extrapolation of motion patterns to block edges, improving both encoding efficiency and reconstructed video quality.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of energy

If motion vectors are handled using conventional methods, then the decoding process can proceed, but bitrate requirements are not optimized

Engineering Contradiction:
ImprovebitrateVSAvoidprediction mechanism complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent segments the motion estimation process into distinct phases: boundary extension with pixel padding, motion vector prediction using extended boundaries, and residual coding. This segmentation allows each phase to be optimized independently, with the boundary extension phase preparing data structures that reduce the bitrate required for transmitting motion information while managing the complexity through modular processing steps.

Inventive Principle:
Principle #1Segmentation

3Measurement precision

If boundary extension techniques are applied, then motion vector prediction accuracy is improved, but the processing time increases

Engineering Contradiction:
Improvemotion vector prediction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The boundary extension and pixel padding are performed as a preliminary action before the main motion estimation loop. By pre-computing the extended boundary pixels and storing them in an expanded reference image buffer, the patent avoids repeated boundary calculations during motion estimation, thereby improving motion vector prediction accuracy while minimizing the time penalty to a single preprocessing pass.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3434019B1An apparatus, a method and a computer program for video coding and decoding
Publication Date: 2026.03.25 NOKIA TECHNOLOGIES OY
  • EP3434019B1 patent drawingFigure 1~2
  • EP3434019B1 patent drawingFigure 3
  • EP3434019B1 patent drawingFigure 4

AI summary

There is provided a method comprising encoding an uncompressed constituent frame into a first encoded picture, said encoding also resulting into a reconstructed first picture and said constituent frame having an effective picture area within the first reconstructed picture, performing either of the following as a part of said encoding: inserting at least one sample value outside the effective picture area to form a boundary extension for the constituent frame in the reconstructed first picture; or saturating or wrapping oversample locations outside the effective picture area to be within the effective picture area. There is also provided a method comprising receiving an encoded picture, decoding the encoded picture to form a reconstructed constituent frame of the picture having an effective picture area;and performing either of the following: filling an area outside the effective picture area to produce a padded reference picture, wherein the filled area forms a boundary extension; or determining that when referring to sample locations outside the effective picture area in decoding, said sample locations are saturated or wrapped over to be within the effective picture area.