Weld Pool Image Segmentation of Hump Formation Based on Fuzzy C-Means and Chan-Vese Model

Abstract

The geometric parameters of the weld pool can reflect welding quality, and thus, it can be important to detect the weld pool contour accurately and reliably. A passive vision system was designed to obtain the weld pool image of hump formation by an infrared transmitting filter, and a method of weld pool image segmentation strategy based on a Chan-Vese (CV) model with fuzzy C-means (FCM) is proposed. The FCM-CV algorithm uses an FCM model to set the initialization contour and then extracts the contour of the hump by the active contour CV model. FCM-CV algorithm eliminates the problem that the CV model is sensitive to the initial contour, and thus, the FCM-CV algorithm can extract the contours of the weld pool under different processing conditions. The contour of the hump is segmented, and the weld pool length feature is extracted. The results show that the truncation of the weld pool length is the main image feature reflecting the formation of the hump.