Image Correction for Accumulated Errors in Stitched Views

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image correction technologies fail to effectively cancel accumulated errors in combined images formed by connecting multiple partial images, leading to curvature and step-like distortions in the combined image, despite previous methods addressing image distortion before connection.

Innovation Solution

An image correction device and method that specifies a correction target area in a combined image, divides it into multiple areas, and converts these areas into rectangles, using vertex or side designations and geometric conversion to eliminate accumulated errors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If multiple partial images are connected to create a combined image, then the field of view is expanded, but accumulated errors cause curvature and distortion in the combined image

Engineering Contradiction:
Improvefield of viewVSAvoidimage accuracy
Core Design Contradiction:
Area of stationary objectVSManufacturing precision

Solution Approach 1:

The invention divides the correction process into segments by identifying individual partial image boundaries within the combined image. Each partial image is processed separately to detect and correct its specific geometric distortion, then reassembled to form the final corrected combined image. This segmentation approach allows precise correction of accumulated errors at each boundary while maintaining the expanded field of view.

Inventive Principle:
Principle #1Segmentation

2Manufacturing precision

If image distortion correction is performed on each partial image before connection, then image accuracy is improved, but accumulated errors at connection boundaries still occur

Engineering Contradiction:
Improveimage accuracyVSAvoiderror cancellation
Core Design Contradiction:
Manufacturing precisionVSReliability

Solution Approach 1:

The invention implements a feedback mechanism by detecting the actual positions of boundary lines between partial images in the combined image, comparing them with expected straight line positions, and using this information to calculate and apply corrective geometric transformations. This feedback loop continuously adjusts the correction parameters to eliminate accumulated errors at connection boundaries, ensuring high reliability in error cancellation.

Inventive Principle:
Principle #23Feedback

3Stability of the object's composition

If rotation variation is restricted during connection, then curvature is suppressed, but step-like distortions appear in frame lines

Engineering Contradiction:
Improvecurvature suppressionVSAvoidframe line accuracy
Core Design Contradiction:
Stability of the object's compositionVSManufacturing precision

Solution Approach 1:

The invention dynamically adjusts multiple parameters including rotation angles, scaling factors, and translation vectors for each partial image based on detected boundary line positions. By allowing controlled rotation variations and using geometric conversion with adjustable parameters, the system eliminates step-like distortions in frame lines while maintaining overall composition stability, achieving both curvature suppression and frame line accuracy.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3605447B1Image correction device, image correction method and program
Publication Date: 2025.09.24 FUJIFILM CORP
  • EP3605447B1 patent drawingFigure 1
  • EP3605447B1 patent drawingFigure 2
  • EP3605447B1 patent drawingFigure 3

AI summary

Provided are an image correction device, an image correction method, and a program capable of canceling accumulated error at the time of connection from a wide field-of-view image in which a plurality of partial images are connected. An image correction device includes an image acquisition unit 22 that acquires a combined image including a plurality of partial images obtained by performing division imaging on a rectangular area in a real space using an imaging device 12, an area specifying unit 34 that specifies a correction target area in the combined image corresponding to the rectangular area in the real space, a division unit 36 that divides the correction target area into a plurality of divided areas, and a geometric conversion unit 40 that converts each of the plurality of divided areas into a rectangle.