Border Region Processing in Image Frame Rate Up-Conversion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional frame rate up-conversion algorithms fail to effectively handle border regions in images, leading to artifacts where black lines can move into the picture and disrupt the background motion, causing unwanted border lines to appear in interpolated or extrapolated images.

Innovation Solution

A method is introduced to identify border regions in images by calculating average and difference values of luminance properties across rows and columns, using a classification parameter to distinguish border regions from internal regions, and excluding border elements from interpolation or extrapolation to prevent these artifacts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional frame rate up-conversion algorithms are used without special border treatment, then the processing is simple and fast, but border lines move into the picture and create artifacts in the interpolated image

Engineering Contradiction:
Improveimage qualityVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The image is divided into border regions and internal regions based on luminance threshold detection. Border regions are identified by comparing luminance values against thresholds and are handled separately from internal regions, preventing border lines from moving into the picture while maintaining simple processing

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different processing methods are applied to different regions: border regions use one processing approach (preserving border lines) while internal regions use another (motion estimation). This local differentiation resolves the contradiction by applying complexity only where needed

Inventive Principle:
Principle #3Local quality

2Reliability

If motion estimation is applied to all regions including borders, then the processing is uniform and simple, but border lines are incorrectly processed and move into the picture

Engineering Contradiction:
Improveborder line stabilityVSAvoidprocessing simplicity
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

Before applying motion estimation, the algorithm preliminarily identifies border regions by detecting luminance thresholds. This preliminary classification allows border lines to be protected from incorrect motion estimation processing, maintaining their stability while keeping the overall process simple

Inventive Principle:
Principle #10Preliminary action

3Productivity

If no special border handling is implemented, then the algorithm remains simple and fast, but artifacts appear where border lines enter the internal region

Engineering Contradiction:
Improveprocessing speedVSAvoidartifacts
Core Design Contradiction:
ProductivityVSObject-generated harmful factors

Solution Approach 1:

Border regions are extracted and separated from the main processing flow. By taking out border regions and handling them separately, the algorithm prevents harmful artifacts while maintaining high processing speed, as the extraction uses simple luminance threshold comparisons

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2108177B1Border region processing in images
Publication Date: 2019.04.10 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP2108177B1 patent drawingFigure 1~13
  • EP2108177B1 patent drawingFigure 2~4
  • EP2108177B1 patent drawingFigure 5~6

AI summary

A border region (70) is identified in an image (40) by calculating an average of pixel values in a row (20) or column (10) of the image (40). Differences in property values are determined between each pixel (30) in the row (20) or column (10) and a neighboring pixel (30) present on a same column (10) or row (20) but in a neighboring row (20) or column (10). An average difference is calculated based on these differences. The pixels (30) in the row (20) or column (10) are classified as belonging to a border region (70) or internal region (75) of the image (40) based on the average pixel value and the average pixel difference.