Weld defects detection for X-ray linear array real-time imaging

A method of the noise reduction, region segmentation and welding defects extraction was proposed accerding to weld image of X-ray linear array detector real-time scanning.The adaptive median filtering was applied to reduce noise, and the weld region was intersected by the variance ratio of interclass and intraclass segmentation with mathematical morphology method, and then the welding defects were extracted by high frequency enhancement.Simulation results show that the adaptive median filter can effectively remove noise while the weld and defects edge detail was reserved.By variance ratio of interclass and intraclass together with mathematical morphology method, the weld metal region can be distinguished accurately from the base metal;the grey changes in weld center can be extruded with the high frequency enhancement filtering, consequently the defects can be detected.