CLASSIFICATION AND DETECTION OF DEFECTS IN TUBULAR PRODUCTS USING MACHINE VISION
Nataliia V. Stelmakh, Ihor Mastenko · Scientific notes of Taurida National V I Vernadsky University Series Technical Sciences · 2024
Inspection is a critical component of a production management system, involving the measurement and inspection of products to ensure that they meet required specifications and performance.Quality control standards are usually dictated by quality assurance approaches such as ISO 9001.These approaches define protocols for the manufacturing process, and quality control ensures that the required tasks are performed.In a manufacturing environment, businesses can perform various types of quality control, including material inspection, in-process inspection, and final inspection.By performing these checks, you can ensure that quality control is on the right track, resulting in lower costs and increased efficiency.Implementing quality control best practices can improve quality control.The key ingredients for success in building a quality control system are automation, inspection, and tracking.Non-automated visual inspection, which is still used in many enterprises, has a number of disadvantages, namely subjectivity and low reliability, as well as low performance on complex products with fine topology.According to the operators, inspection of complex products using simple optical devices is very stressful on their eyes, which is probably a danger to their health. The growing interest in computer vision technology is due to the non-contact method of control, high accuracy and productivity, at a relatively low costThe paper considers the problem of controlling the output characteristics of finished products, which is proposed to be solved using computer vision methods, and proposes a mathematical description of one type of defect using the Hough transform, which is successfully used to describe and detect objects of round shape.Also, the paper analyses the most typical defects for the object of control of a tubular body and proposes a classification of these defects by grades for further use in training a neural network.