Image Processing Sensor Threshold Setting via Background Subtraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing sensors face challenges in stabilizing the separation of non-defective and defective products due to the complexity of registering model and non-model images, leading to difficulties in determining accurate thresholds and user-friendly operation.

Innovation Solution

The image processing sensor allows for easy setting of thresholds by enabling two-point or three-point registration modes, where users can register non-defective, defective, and background images separately, using a user-friendly interface with adjustable display magnification and matching degree thresholds, reducing the influence of background elements and improving defect detection accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If only model image registration is performed, then the setting operation is simple, but the threshold cannot be stably determined to separate non-defective and defective products

Engineering Contradiction:
Improvesetting operation simplicityVSAvoidthreshold determination accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent segments the teaching process into multiple independent registration modes (one-point, two-point, three-point teaching) that can be selected based on specific needs. Each mode registers different combinations of images (model image, non-model image, background image) to calculate thresholds, allowing users to choose the appropriate level of complexity for their application while ensuring accurate defect detection.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If both model image and non-model image are registered, then the threshold determination improves, but the operation and registration procedure becomes complicated

Engineering Contradiction:
Improvethreshold determination accuracyVSAvoidregistration procedure complexity
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The patent implements dynamic operation modes that automatically adjust the registration procedure based on the selected teaching mode. The system dynamically determines which images to register and how to calculate thresholds, adapting the complexity of the procedure to match the user's needs while maintaining accurate defect detection capabilities.

Inventive Principle:
Principle #15Dynamics

3Ease of operation

If background elements are included in the model image, then the registration is simpler, but the defect detection accuracy deteriorates due to background influence

Engineering Contradiction:
Improveregistration simplicityVSAvoiddefect detection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent extracts background elements from the imaging target through dedicated background image registration in two-point and three-point teaching modes. By separately registering the background image and subtracting it from the model image, the system removes background interference while maintaining registration simplicity, thereby improving defect detection accuracy without significantly increasing operational complexity.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3258442B1Image processing sensor, image processing method, image processing program, and computer-readable recording medium having image processing program recorded therein
Publication Date: 2024.04.10 KEYENCE CORP
  • EP3258442B1 patent drawingFigure 1A~1B
  • EP3258442B1 patent drawingFigure 2A~2B
  • EP3258442B1 patent drawingFigure 3A~3C

AI summary

An image processing sensor includes an image registering unit configured to register a non-defective product image, a defective product image, and a background image, a differential-image generating unit configured to generate a first differential image from the non-defective product image and the background image and generate a second differential image from the defective product image and the background image; a matching-degree calculating unit configured to calculate a matching degree indicating a degree of feature matching of the second differential image with respect to a model image corresponding to the first differential image, and a threshold calculating unit configured to calculate, on the basis of the matching degree, a threshold used in an operation mode.