Comparative study of image processing algorithms to detect defects in cast components

Authors

Muralidhar, Nikith

Issue Date

2023

Degree

Publisher

DBS Library Press

Rights

Items in eSource are protected by copyright. Previously published items are made available in accordance with the copyright policy of the publisher/copyright holder.

Abstract

In the manufacturing industry, the non-destructive evaluation (NDE) of components is crucial. These cast components are susceptible to blowholes and other anomalies. If such flaws are included in the components, the fatigue life will be harmed, which would almost certainly result in catastrophic accidents. Humans currently evaluate cast components by various methods. We propose an automatic approach for detecting faults in casts with the goal of producing a category that will eliminate the need for manual testing. The technique looks for defects in cast components, In the previous years, image processing technology has advanced significantly. The method proposed utilizes Convolutional Neural Networks (CNN) and Support Vector Classifiers (SVC’s). This process classifies if the component has a defect or not. According to the hypothesis, human examiners may benefit from the approach because it reduces their workload.