UROP Openings

Automating metallographic image analysis

Project Description

Welding is a fabrication process whereby two or more parts are fused together by means of heat, pressure or both forming a join as the parts cool. Welding is usually used on metals and thermoplastics. A large database of weld images (weld cross-sections) from an industry partner will be used to help to explore and create machine vision and machine learning algorithm which automatically measure features like weld penetration, width, defects, etc. The focus will be on learned feature recognition, segmentation, and dimensioning and results in models of images with reference points, feature dimensions, and contours. Impact: This project is of interest to many manufacturers and applied to defect detection in welding, incoming material inspection, and in-line measurement. The approaches can be applied to other fields, for example, in medical radiography measuring the size of objects of interest (e.g. cancerous growth, etc.).

Pre-requisites

Experience and/or interest in machine vision and machine learning algorithms