A Machine Vision System for Detecting Automotive Gear Defects
Loading...
Date
Authors
Allam, Abdelrahman
Journal Title
Journal ISSN
Volume Title
Publisher
University of Guelph
Abstract
This thesis presents new methods that can be used for the automated quality inspection of automotive gears. Typically, gears are manually inspected at the end of the manufacturing process by trained human inspectors. This process is expensive, somewhat subjective, and suffers from errors due to human fatigue. Automating this process would improve quality and traceability. This would result in fewer rejected gears, thus reducing the overall costs. This thesis investigates multiple approaches ranging from simple image processing techniques to deep learning. The best results were obtained using a hybrid system that combines deep learning with domain knowledge.
Description
Keywords
Automotive Gears Inspection, Machine Vision Inspection, Gear Defect Detection, Visual Inspection Using Image Processing, Visual Inspection Using Deep Learning, Visual Inspection Using Machine Learning