Comparison of algorithms to quantify the damaged area in CFRP ultrasonic scans

CFRP structures can be damaged in different ways during service. Therefore, the effect of a damage on the structural behaviour has to be considered during the design process. Often the damage size and shape are used to characterize damages. So a precise knowledge of the size and shape of damaged areas is essential. In the present study a comparison of five different algorithms determining the damaged area size and shape of a CFRP specimen ultrasonic C-scan is presented. The considered algorithms are a pixel count algorithm, a Convex Hull algorithm, a Support Vector Machine algorithm, a Snake algorithm and a combination of two of them. Their reliability, efficiency and robustness are tested and compared by applying the algorithms to real world ultrasonic C-scans and calculating the damage size and shape automatically.