Computer vision techniques for wear debris analysis (CVTWDA)

Abstract

Computer vision is a process to locate and recognise objects in digital images. The terms "wear particle" or "wear debris" come from the field of "tribology". Tribology is the study of wear, friction and lubrication. Wear particles are produced in all machines containing moving parts in contact. This paper addresses the analysis of microscopic wear particles generated by wear mechanisms by using computer vision and image processing techniques. The aim is to develop an automated system to classify and identify wear particles in terms of their morphological and textural features and by using the information obtained to predict wear failure modes in engines and other machinery. An interactive control system, CAVE, is devised to process and store the quantitative information of particle size, shape and edge details. The identification and analysis of wear particles are effectively demonstrated by the devised system with the ability to make human-like, even better and objective diagnosis. This obviates the need for specialists and reliance on human visual inspection techniques.