Edge-Guided Demosaicing for Sharper Images and Fewer Color Artifacts

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing demosaicing methods in image sensors result in artifacts such as zipper effect, block noise, reduced image detail, false color, and blurring due to interpolation errors, particularly at edges, leading to a trade-off between noise filtering and sharpness.

Innovation Solution

A demosaicing method and device that includes a main edge estimator, a main edge enhancer, and a sub interpolator to improve edge detection and directionality, using high-frequency edge enhancement and gradient-based interpolation to reduce color errors and enhance image sharpness.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If linear interpolation is used to restore missing color pixel values, then the demosaicing process is simple and fast, but image sharpness decreases and edges become blurred

Engineering Contradiction:
Improvedemosaicing process simplicityVSAvoidimage sharpness
Core Design Contradiction:
Ease of operationVSManufacturing precision

Solution Approach 1:

The patent changes the parameter of interpolation method from simple linear interpolation to gradient-based directional interpolation. By calculating gradients and determining edge directions, the algorithm adapts the interpolation approach based on local image characteristics, preserving edge sharpness while maintaining computational feasibility

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces dynamic adaptation by estimating edge directions and adjusting interpolation strategies based on detected edge orientations. The algorithm dynamically selects interpolation directions perpendicular to edges, making the demosaicing process adaptive rather than static, thereby preserving image sharpness

Inventive Principle:
Principle #15Dynamics

2Object-affected harmful factors

If filtering is applied to reduce noise, then noise is reduced, but sharp edges become blurred

Engineering Contradiction:
Improvenoise reductionVSAvoidedge sharpness
Core Design Contradiction:
Object-affected harmful factorsVSManufacturing precision

Solution Approach 1:

The patent applies different processing strategies to different local regions of the image. By detecting edge directions locally, the algorithm performs directional interpolation that preserves edge sharpness in edge regions while applying appropriate filtering in non-edge regions, achieving local optimization rather than uniform processing

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the image processing into distinct stages: gradient calculation, edge direction estimation, and directional interpolation. This segmentation allows the algorithm to selectively apply noise reduction and edge preservation techniques to different components, avoiding the trade-off between noise filtering and edge sharpness

Inventive Principle:
Principle #1Segmentation

3Productivity

If color channels are restored in sequence using previous step results, then the processing is efficient, but errors accumulate

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidcolor accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary gradient calculations and edge direction estimations using the green channel data before interpolating red and blue channels. This preliminary action provides accurate directional information that guides the subsequent color channel restoration, preventing error accumulation by establishing a reliable foundation first

Inventive Principle:
Principle #10Preliminary action

4Ease of operation

If averaging of pixels crossing edges is performed, then interpolation is simple, but image sharpness at edges decreases

Engineering Contradiction:
Improveinterpolation simplicityVSAvoidedge sharpness
Core Design Contradiction:
Ease of operationVSManufacturing precision

Solution Approach 1:

Instead of averaging pixels across edges (which blurs edges), the patent inverts the approach by identifying edge directions first and then performing interpolation perpendicular to edges. This inversion prevents averaging across edge boundaries, thereby preserving edge sharpness while maintaining interpolation functionality

Inventive Principle:
Principle #13The other way round (Inversion)

Data Source

PatentEP4386656B1Demosaicing method and demosaicing device
Publication Date: 2026.05.06 SAMSUNG ELECTRONICS CO LTD
  • EP4386656B1 patent drawingFigure 1
  • EP4386656B1 patent drawingFigure 2
  • EP4386656B1 patent drawingFigure 3

AI summary

A demosaicing method for an input mosaic image includes: obtaining an input image, for each sub target pixel, estimating a main edge direction based on main color pixel values; determining the main edge direction by performing edge enhancement based on the input mosaic image; generating a main color demosaiced image by performing a main interpolation based on the main edge direction to generate interpolated main color pixel values corresponding to the sub target pixels; and generating a first sub color demosaiced image and a second sub color demosaiced image by performing a sub interpolation to generate interpolated first sub color pixel values and interpolated second sub color pixel values for the sub target pixels and for main target pixels corresponding to the main color pixel values.