Metal Ball Surface Defect Inspection via Multi-Angle Image Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional metal ball sorting machines lack the capability to provide accurate analysis data to improve the manufacturing process, as they only sort metal balls into acceptable and unacceptable categories without detailed defect analysis.
Innovation Solution
A method utilizing image recognition that involves feeding metal balls onto a rotary disc with light-transmitting material and light source units, capturing images of both surfaces, and comparing them against a database to identify and classify defects, allowing for precise sorting and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional light-based sorting machines are used to sort metal balls into acceptable and unacceptable categories, then the sorting function is achieved, but detailed defect analysis and classification capability is lost
Solution Approach 1:
The inspection system segments the metal ball surface into multiple regions by capturing images at different rotation angles. The rotary disc divides the inspection process into multiple angular positions, allowing each surface region to be examined separately and classified into different defect types based on its specific characteristics.
Solution Approach 2:
The system transitions from simple binary classification (acceptable/unacceptable) to multi-dimensional defect classification by capturing images at multiple rotation angles. This adds the dimension of angular position and surface orientation, enabling detailed defect type identification such as scratches, strains, and other surface anomalies.
2Measurement precision
If single-surface image capture is used, then the inspection process is simple, but comprehensive defect detection capability is insufficient
Solution Approach 1:
The rotary disc enables continuous rotation of the metal ball through the inspection area, maintaining constant contact between the ball and the imaging system. This continuous action allows multiple surfaces to be inspected sequentially without interrupting the flow, achieving comprehensive coverage while maintaining efficient inspection timing.
Solution Approach 2:
The system captures images at multiple rotation angles before final classification is determined. By preliminarily capturing surface features at various orientations, the system builds a comprehensive defect profile that enables accurate classification, ensuring no surface region is overlooked.
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
Enables accurate sorting and classification of metal balls based on defects, providing detailed analysis data to enhance the manufacturing process by distinguishing between different types of defects, such as scratches and strains, and directing unacceptable balls to different discharge paths.
Implementation Method 1
a light source unit, the rotary disc being formed with plural holes for accommodation of the metal ball, the metal ball having a first surface which is visible from the holes, the light source unit being located at one side of the rotary disc to enable the first surface of the metal ball to produce uniform brightness
Data Source
AI summary
A method for inspecting surface defect of metal balls by image recognition includes the steps of feeding metal ball, capturing image for a first time, making metal ball rotate, capturing image for a second time, comparing images, and discharging metal balls. With the above steps, not only can the metal balls be sorted into the acceptable and the unacceptable metal balls, but the unacceptable metal balls can be sorted into different kinds according to the defects such as scratch, strain, and so on. Hence, effective data can be offered to improve the metal ball manufacturing process accurately.