Image Data Generating Apparatus for Product Inspection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image inspection technologies face challenges in generating suitable inspection images due to variations in brightness conditions and the presence of defects, which can hinder accurate defect detection in product images.

Innovation Solution

An image data generating apparatus using a pre-trained model that performs data compression and decompression to output defect-free images, combined with luminosity adjustment processes to create inspection image data by comparing adjusted input and output images, facilitating effective external appearance inspection of products.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If conventional image inspection methods are used to identify standard pattern position and orientation, then the inspection process can be automated, but the inspection image quality deteriorates when brightness conditions deviate from reference conditions or when defects are present in the standard pattern area

Engineering Contradiction:
Improveautomation of inspection processVSAvoidinspection image quality
Core Design Contradiction:
Extent of automationVSMeasurement precision

Solution Approach 1:

The patent extracts and removes the standard pattern from the image processing pipeline. Instead of relying on standard pattern recognition, the system directly captures and processes the inspection target image, eliminating the source of error when defects occur in standard pattern areas or when brightness conditions vary.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies luminosity adjustment to normalize brightness conditions in the captured image. By adjusting the luminosity parameter of the image data, the system compensates for variations in ambient light and camera settings, ensuring consistent inspection quality across different shooting conditions.

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If the inspection image is captured under varying brightness conditions, then the inspection can be performed in different environments, but the accuracy of defect detection deteriorates due to brightness deviations from reference conditions

Engineering Contradiction:
Improveinspection environment flexibilityVSAvoiddefect detection accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent dynamically adjusts the luminosity parameter of the captured image to match reference brightness conditions. This parameter transformation allows the system to maintain high defect detection accuracy across varying environmental lighting conditions, preserving both adaptability and precision.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If the standard pattern area contains defects, then the inspection target can still be processed, but the position and orientation correction becomes inaccurate, leading to poor inspection image quality

Engineering Contradiction:
Improveinspection throughputVSAvoidposition and orientation accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent eliminates dependency on standard pattern recognition by removing this step from the inspection workflow. The system directly processes the inspection target image without attempting to identify or correct based on standard patterns, thereby avoiding accuracy degradation when defects are present in standard pattern areas.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS11107210B2Image data generating apparatus generating image data for inspecting external appearance of product
Publication Date: 2021.08.31 BROTHER KOGYO KK
  • US11107210B2 patent drawing
  • US11107210B2 patent drawing
  • US11107210B2 patent drawing

AI summary

In an image data generating apparatus, an image processor acquires target output image data converted from target input image data by a pre-trained model. The target input image data represents a captured image of a target product. The target output image data represents an image of a target product free from defects. The pre-trained model performs data compression and data decompression on inputted image data. The image processor generates first data by performing a first generation process on the target input image data. The first generation process includes a luminosity adjustment process adjusting luminosity in image data. The image processor generates second data by performing a second generation process including the luminosity adjustment process on the target output image data. The image processor generates inspection image data by using a difference between the first data and the second data for inspecting an external appearance of the target product.