Weld defect detection based on Gaussian curve
暂无分享,去创建一个
Develops a weld defect detection methodology based on the assumption that a line profile of a defectless weld image can be approximated by a Gaussian distribution curve. The line profile variations of a weld image caused by defects are classified into three defect patterns, defect-peaks, defect-troughs and defect-slant-concaves. Dark image enhancement is used to control the level of the noises which otherwise would have become worse in normalization. Two kinds of B-spline curve fittings, tight fitting and loose fitting, are performed to facilitate defect identification. The purpose of tight fitting is to reduce the noises but keep the profile variations caused by defects, while that of loose fitting is to restore the bell shape as if no defects would have occurred. The roughness of a line image profile is defined and used to estimate the smoothing factor used for fitting the line profile. The results of preliminary tests showed that more than 90% of defects are successfully detected.
[1] Paul Dierckx,et al. Curve and surface fitting with splines , 1994, Monographs on numerical analysis.
[2] P. Rose,et al. Automatic recognition of weld defects in x-ray inspection , 1992 .
[3] Norbert Meyendorf,et al. USE OF AUTOMATIC IMAGE PROCESSING FOR MONITORING OF WELDING PROCESSES AND WELD INSPECTION , 1989 .
[4] A Gayer,et al. Automatic recognition of welding defects in real-time radiography , 1990 .