Image Processing Sensor Threshold Calculation for Defect Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image processing sensors face challenges in accurately distinguishing between non-defective and defective products due to background elements and illumination environments, leading to deteriorated defect detection accuracy, as they rely solely on model image registration without the ability to register non-model images.

Innovation Solution

An image processing sensor system that includes an imaging unit, a display unit, an operation unit, a threshold calculating unit, and a display control unit, allowing for the registration of both non-defective and defective product images, with live and still image displays on separate screens to clearly differentiate features, thereby stabilizing defect detection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If only model image registration is performed, then the system is simple to operate, but defect detection accuracy deteriorates due to background elements and illumination effects

Engineering Contradiction:

Improveease of operationVSAvoiddefect detection accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent segments the image registration process into two distinct modes: model image registration and non-model image registration. This segmentation allows the system to handle different types of images with different processing requirements, thereby improving defect detection accuracy while maintaining ease of operation through automated threshold calculation.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a threshold calculation unit as an intermediary that automatically calculates the threshold for pass/fail determination based on both model and non-model images. This intermediary component resolves the contradiction by eliminating the need for manual threshold setting, thus improving accuracy without complicating the operation.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Device complexity

If threshold is calculated only from model image, then the process is simple, but it is difficult to determine a threshold that can stably separate non-defective and defective products

Engineering Contradiction:

Improveprocess complexityVSAvoidthreshold stability

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent merges the information from both model images and non-model images to calculate the threshold. By combining multiple image sources, the threshold calculation becomes more robust and stable, reliably separating non-defective and defective products without significantly increasing process complexity.

Inventive Principle:

Principle #5Merging (Combining)

3Ease of manufacture

If background elements are included in model image, then the registration process is simplified, but defect detection accuracy deteriorates

Engineering Contradiction:

Improveregistration simplicityVSAvoiddefect detection accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent extracts and separates the background elements from the product features by introducing non-model image registration. This extraction process allows the system to focus on product-specific features while eliminating the negative impact of background elements and illumination effects on defect detection accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

4Ease of operation

If only one image type is registered, then the system is easier to operate, but the ability to distinguish differences between non-defective and defective products is reduced

Engineering Contradiction:

Improveoperation simplicityVSAvoiddifference detection capability

Core Design Contradiction:

Ease of operationVSLoss of information

Solution Approach 1:

The patent implements a dynamic image registration system that can adaptively select and register different types of images (model and non-model) based on the inspection requirements. This dynamic approach preserves all relevant information for difference detection while maintaining ease of operation through automated processing.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10096103B2Image processing sensor, image processing method, image processing program, and computer-readable recording medium and device having image processing program recorded therein

Publication Date: 2018.10.09 KEYENCE CORP

AI summary

An image processing sensor includes a threshold calculating unit configured to calculate a threshold with respect to a matching degree indicating a degree of feature matching of a first image including an inspection target object that should be distinguished as a non-defective product and a second image not including the inspection target object that should be distinguished as the non-defective product and a display control unit configured to cause a display unit to display the first image and the second image as images used for the calculation of the threshold by the threshold calculating unit. The display control unit causes the display unit to display, on a first registration screen, as a live image, the one image acquired by the imaging unit, and causes the display unit to display, on a second registration screen, as the live image, the other image acquired by the imaging unit and, as a still image, the already registered one image.