Image Error Mapping for Super-Resolution Reconstruction Quality

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image reconstruction techniques introduce errors and artifacts, leading to low image quality and misinterpretation of information, particularly in super-resolution imaging, which current quality assessments struggle to address effectively.

Innovation Solution

The use of a pixel-level analysis of error locations (PANEL) framework, combining rolling Fourier ring correlation (rFRC) and a reference map, to generate a target map that quantitatively maps errors in images without requiring a ground-truth reference, facilitating robust evaluation of image quality across various imaging modalities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If image reconstruction techniques are used to generate images from acquired data, then image resolution and detail can be improved, but image errors and artifacts are introduced leading to reduced image quality

Engineering Contradiction:
Improveimage resolutionVSAvoidimage quality
Core Design Contradiction:
Manufacturing precisionVSReliability

Solution Approach 1:

The image is divided into multiple local regions or patches, and error mapping is performed independently for each region. This segmentation allows the system to identify and correct local artifacts while preserving global image structure, resolving the contradiction between achieving high resolution through reconstruction and maintaining overall image quality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

An error map is introduced as an intermediary component between the reconstructed image and the final output. This error map quantifies reconstruction artifacts and guides corrective processing, enabling the system to maintain high resolution while eliminating quality-degrading errors through the intermediary error assessment layer.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If conventional error assessment methods are used, then some image quality metrics can be obtained, but they fail to provide precise error location mapping particularly for super-resolution imaging

Engineering Contradiction:
Improveerror assessment accuracyVSAvoidassessment method complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The error assessment transitions from global metric evaluation to pixel-level spatial mapping by introducing a second dimension of detail. This transforms the error assessment from aggregate quality scores to location-specific error maps, enabling precise identification of artifacts while maintaining manageable computational complexity through efficient algorithms.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Measurement precision

If ground-truth reference images are used for error mapping, then error quantification can be achieved, but the method becomes inapplicable when ground-truth is unavailable

Engineering Contradiction:
Improveerror quantificationVSAvoidmethod applicability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The error mapping system performs self-assessment by comparing different reconstructed versions or using intrinsic image properties rather than requiring external ground-truth references. This self-service capability enables the method to quantify errors and identify artifacts autonomously, making it universally applicable to various imaging scenarios including super-resolution where ground-truth is typically unavailable.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4396767B1Systems and methods for image processing
Publication Date: 2026.01.07 PEKING UNIV
  • EP4396767B1 patent drawingFigure 1
  • EP4396767B1 patent drawingFigure 2~3
  • EP4396767B1 patent drawingFigure 4

AI summary

Systems and methods for image processing are provided in the present disclosure. The systems may obtain a first image and a second image associated with a same object; determine a plurality of first blocks of the first image and a plurality of second blocks of the second image, the plurality of second blocks and the plurality of first blocks being in one-to-one correspondence; determine a plurality of first characteristic values based on the plurality of first blocks and the plurality of second blocks; and/or generate a first target map associated with the first image and the second image based on the plurality of first characteristic values.