Spectroscope-Based Visual Inspection for Foreign Matter Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual inspection methods, such as those using polarizing plates and image recognition technology, face challenges in accurately detecting foreign matters due to operator skill variability and neglect of wavelength components, leading to inconsistent results.
Innovation Solution
A visual inspection device equipped with a spectroscope and camera that captures spectral images based on an inspection wavelength, calculates light intensity changes, and generates a luminance image for enhanced detection, allowing for accurate identification of foreign matters without requiring operator skill.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If image recognition technology is used to detect foreign matter based on luminance distribution, then automation is improved, but measurement precision deteriorates because wavelength components are not considered
Solution Approach 1:
The patent changes the parameter of wavelength selection by using a spectroscope to identify the inspection wavelength corresponding to the target object, and then captures spectral images at this specific wavelength. This parameter change enables both automated operation and high detection precision by focusing on the most discriminative wavelength for detecting foreign matter.
2Ease of operation
If polarizing plates are used in microscopes to enhance contrast, then ease of operation is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential function of wavelength selection from the complex polarizing plate system. By using a spectroscope to select the inspection wavelength and capturing spectral images, the system achieves foreign matter detection without requiring polarizing plates, thereby reducing device complexity while maintaining ease of operation.
3Measurement precision
If spectral images are captured and analyzed, then measurement precision is improved, but loss of time increases due to data handling
Solution Approach 1:
The patent extracts only the essential information from spectral images by calculating the light intensity change amount between the spectral image and reference image, and then calculating a parameter indicating variation in the light intensity change amount. This extraction of key parameters reduces data handling time while maintaining high detection precision.
Solution Approach 2:
The patent performs preliminary action by pre-calculating the inspection wavelength based on spectral measurement information before capturing spectral images. This preliminary wavelength selection optimizes the inspection process and reduces processing time during actual detection.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The device enables high-accuracy detection of foreign matters by analyzing spectral images, reducing data handling and improving processing speed, and allows operators to easily identify anomalies through enhanced luminance images.
Implementation Method 1
a spectroscope that selectively transmits a wavelength of light reflected by a target object
Data Source
AI summary
A visual inspection device includes a spectroscope that selectively transmits a wavelength of light reflected by a target object, a camera that receives the light transmitted through the spectroscope and captures a spectral image of the target object, and one or a plurality of processors, in which the one or the plurality of processors are programmed to execute a method including: controlling the spectroscope to obtain an inspection wavelength corresponding to the target object, and acquiring, from the camera, the spectral image based on the inspection wavelength, calculating a light intensity change amount that is a difference between a light intensity of each pixel of the spectral image and a predetermined reference value, calculating a parameter indicating variation in the light intensity change amount in a pixel range including a target pixel of the spectral image, and determining whether or not the parameter is within a predetermined range.