Gömülü sistem tabanlı bir hatalı ürün tespit sistemi
Loading...
Date
2019-04-03
Authors
Bayram, Raif Bayrak
Journal Title
Journal ISSN
Volume Title
Publisher
Bursa Uludağ Üniversitesi
Abstract
Bu çalışmada, metal sektöründe kullanılmak üzere gömülü sistem tabanlı bir hatalı ürün tespit sistemi geliştirilmiştir. Ürün grubu olarak sektörde sıklıkla üretilen ve dairesel boşluklar içeren sac levhalar seçilmiştir. Levhalar üzerindeki dairesel boşluklara ait bilgiler dairesel Hough dönüşümü kullanılarak elde edilmiştir. Hatalı ürünlerin tespiti referans görüntü kullanılarak karşılaştırma yoluyla yapılmıştır. Gömülü sistem olarak Raspberry Pi Model 3 B+ seçilmiştir. Geliştirilen sistemin başarım incelemesi çalışmaya özel oluşturulmuş farklı çözünürlükteki görüntü kümeleri üzerinde yapılmıştır. Yapılan deneyler sonucunda geliştirilen sistemin 10MP çözünürlükteki görüntü kümesinde %96,29 doğruluk oranına sahip olduğu gösterilmiştir.
In this work, an embedded system based product defect detection system has been developed for the metal industry. As a test product, sheet metal plates which are used very often in the metal industry containing circular holes have been used. The geometrical information about the holes on the plates have been obtained using Hough circular transformation. Detection of defective products was made by comparison using the reference image. As an embedded system Raspberry Pi Model 3 B+ have been used. Performance analysis of the developed system have been carried out on a special image sets having different resolutions which are obtained for the study. Based on the experimental results it has been shown that the developed system reached 96.29% accuracy rate for the image set with 10MP resolution.
In this work, an embedded system based product defect detection system has been developed for the metal industry. As a test product, sheet metal plates which are used very often in the metal industry containing circular holes have been used. The geometrical information about the holes on the plates have been obtained using Hough circular transformation. Detection of defective products was made by comparison using the reference image. As an embedded system Raspberry Pi Model 3 B+ have been used. Performance analysis of the developed system have been carried out on a special image sets having different resolutions which are obtained for the study. Based on the experimental results it has been shown that the developed system reached 96.29% accuracy rate for the image set with 10MP resolution.
Description
Keywords
Ürün hata tespiti, Görüntü işleme, Gömülü sistem, Hough dönüşümü, Product defect detection, Image processing, Embedded system, Hough transform
Citation
Bayram, R. B. ve Yılmaz, E. (2019). "Gömülü sistem tabanlı bir hatalı ürün tespit sistemi". Uludağ Üniversitesi Mühendislik Fakültesi Dergisi, 24(1), 391-400.