摘要 : This article describes an algorithm used to automatically detect differential coil eddy current (EC) defect signal features using a sum of Gaussians approach. Defect signals are decomposed into positive and negative portions then ... 展开 This article describes an algorithm used to automatically detect differential coil eddy current (EC) defect signal features using a sum of Gaussians approach. Defect signals are decomposed into positive and negative portions then a thresholded binary image is constructed using the Parzen–Rosenblatt window technique. Mathematical morphology operators are then used to denoise the thresholded binary image. Peaks are detected in the denoised binary images using a simplified variant of the <inline-formula> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula>-means algorithm. The resulting parameters obtained are input into a constrained optimization algorithm to obtain an estimate of the denoised EC defect signal using a minimum summed-square error objective function. The algorithm performance is investigated over a range of electrostatically discharged machine (EDM) notch return signal windows that vary in coil diameter, test material, test frequency, and number of defects in the investigated region. 收起