Neural Image Inspection for High-Speed Production Quality Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current quality control methods for mass-produced products, such as food packaging, are inefficient, slow, inaccurate, and prone to human error, requiring manual visual inspection to detect faulty products.
Innovation Solution
A fully autonomous quality control system utilizing neural deep learning algorithms and image processing techniques to inspect products on a production line, eliminating the need for human intervention by analyzing images of products to detect morphological, chromatic, and aesthetic defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual visual inspection is used for quality control, then human operators can detect faulty products, but the process becomes inefficient, slow, and prone to human error
Solution Approach 1:
The patent replaces the manual visual inspection system with an automated image processing system that captures images of products, segments them into multiple images, and analyzes them using a processor. This substitution eliminates human operators from the inspection process, thereby removing human error while simultaneously increasing inspection speed through automated high-speed imaging and processing capabilities.
2Productivity
If automated image processing is implemented, then inspection speed and accuracy improve, but system complexity increases
Solution Approach 1:
The patent divides a single captured image into multiple segmented images, each representing different portions or aspects of the original product image. This segmentation allows the system to analyze specific features independently and in parallel, improving inspection efficiency and accuracy while managing system complexity through modular processing of divided image data rather than analyzing one complete image at once.
Data Source
Figure 1~2
Figure 3~4
Figure 5~7
AI summary
A quality control system (100) includes: a conveyor (10) on which parts (2) to be inspected are arranged, image acquisition means (3) suitable for acquiring images (I) of the parts (2) on the conveyor, and a control unit (4) suitable for receiving and processing the images (I) acquired by the image acquisition means (3). The control unit (4) has an inspection program (5) and a control program (6) which are based on a neural network. The inspection program (5) is configured to calculate and store quantities and threshold limits that will be used in the control program (6), and the control program (6) is configured to determine whether each part (2) is compliant or to be rejected.