Smoothing Interpolation Filter for Video Encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video encoding methods face challenges in achieving high-quality image interpolation for accurate inter prediction, particularly in removing high-frequency components that reduce encoding efficiency.

Innovation Solution

A video encoding method using a smoothing interpolation filter to generate sub-pel-unit pixels by interpolating integer-pel-unit pixels, which selects appropriate interpolation filters based on location and smoothness to reduce high-frequency components and improve encoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional interpolation methods are used to generate sub-pel-unit pixels, then inter prediction can be performed, but high-frequency components are not effectively removed reducing encoding efficiency

Engineering Contradiction:
Improveinter prediction accuracyVSAvoidencoding efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies parameter changes by modifying the interpolation filter parameters based on the gradient magnitude of the image blocks. When gradient magnitude exceeds a threshold, a smoothing filter is applied to reduce high-frequency components. This dynamically adjusts the interpolation parameters to balance prediction accuracy and encoding efficiency, resolving the contradiction between maintaining precision and improving productivity.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If smoothing interpolation filter is applied to reduce high-frequency components, then encoding efficiency is improved, but image quality may be reduced

Engineering Contradiction:
Improveencoding efficiencyVSAvoidimage quality
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent implements local quality by applying smoothing interpolation selectively based on local image characteristics. The gradient magnitude calculation identifies regions where smoothing is beneficial, and the filter is applied only to those specific blocks. This localized approach maintains encoding efficiency improvements while preserving image quality in regions where smoothing is not needed, resolving the contradiction between productivity and manufacturing precision.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3276961B1Method and device for interpolating images by using a smoothing interpolation filter
Publication Date: 2024.08.28 SAMSUNG ELECTRONICS CO LTD
  • EP3276961B1 patent drawingFigure 1~2
  • EP3276961B1 patent drawingFigure 3~4B
  • EP3276961B1 patent drawingFigure 4C~5

AI summary

Provided is a method of interpolating an image by determining interpolation filter coefficients. A method of interpolating an image in consideration of smoothing includes differently selecting an interpolation filter based on a sub-pel-unit interpolation location and a smoothness from among interpolation filters for generating at least one sub-pel-unit pixel value located between integer-pel-unit pixels; and generating the at least one sub-pel-unit pixel value by interpolating pixel values of the integer-pel-unit pixels by using the selected interpolation filter.