Polymer weld characterization and defect detection through advanced image processing techniques

Welding defect detection in a radiographic image is vital in industrial non-destructive testing. It is significant in evaluating weld anomalies and surface and subsurface imperfections in welded joints. Digital image processing techniques can make automation feasible in the weld microscopic image interpretation, thus reducing the instances of observational human errors in weld inspection. This technique will give more reliability, speed and reproducibility to the inspection system. This paper uses MATLAB image processing tools for weld defect detection using scanning electron microscope images of ultrasonically welded polymer samples. Image processing features of gray scaling, image resizing, histogram equalization, edge detection, thresholding, filtering, texture analysis, and image segmentation have been implemented for the detection and characterization of defects in weld scanning electron microscope images. Thorough insights into the structure of these defects are an essential step in appreciating the weld's quality. The approach is well-suited for defect detection of any welding technique.