Gradient-Based Image Filtering for HDR Motion Artefact Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for combining images to form High Dynamic Range (HDR) images struggle with handling motion, leading to noticeable artefacts such as ghosting effects, especially on mobile devices with limited processing power.

Innovation Solution

The method involves determining image gradient indications and filter costs for image regions, which are then used to apply filtering operations that depend on image gradients. This approach allows for denoising and blending of images while smoothing transitions and handling motion by treating motion regions as 'inconsistent' areas.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If complex motion tracking procedures are used to handle motion in HDR image processing, then motion artefacts are reduced, but processing time and computational complexity increase significantly

Engineering Contradiction:
Improvemotion handling qualityVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts and utilizes image gradient information to identify motion regions without performing complex motion tracking. By separating the task into gradient computation (simple) and motion region identification (complex), the solution achieves effective motion handling while keeping processing complexity manageable for mobile devices

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent employs a simplified gradient-based approach that computes lightweight gradient indications rather than expensive motion tracking algorithms. This disposable-like approach uses minimal computational resources to achieve the desired motion artifact reduction

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

2Shape

If gradient-based filtering is applied to smooth transitions, then transition sharpness is reduced, but processing time increases

Engineering Contradiction:
Improvetransition smoothnessVSAvoidprocessing time
Core Design Contradiction:
ShapeVSLoss of time

Solution Approach 1:

The patent applies gradient-based filtering selectively only in regions where transitions are detected, rather than processing the entire image uniformly. This local quality approach reduces overall processing time while maintaining transition smoothness where needed

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses a simplified gradient computation that performs partial filtering action focused on transition regions. By applying filtering only where gradients indicate transition zones rather than everywhere, the solution achieves sufficient smoothness with reduced processing time

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3696769B1Image filtering based on image gradients
Publication Date: 2025.04.23 IMAGINATION TECH LTD
  • EP3696769B1 patent drawingFigure 1
  • EP3696769B1 patent drawingFigure 2
  • EP3696769B1 patent drawingFigure 3

AI summary

An image processing method comprising: determining image gradient indications (S312) for at least one image; determining filter costs (S314) for image regions based on the determined image gradient indications for the at least one image; processing data relating to the at least one image including applying a filtering operation (S316) using the determined filter costs for the image regions, so that the processing is performed for image regions in dependence upon image gradients of the at least one image, the processing comprising using the filtering operation to apply denoising to the at least one image such that the extent of the denoising applied to an image region is dependent upon the filter cost for the image region.