A Step Towards Automatic Defect Pattern Analysis and Evaluation in Industrial Radiography using Digital Image Processing

摘要

Industrial radiography is a traditional technique for the identification andrnevaluation of flaws, or defects, such as cracks, porosity and foreignrninclusions found in casting and machined parts. In the past 40 yearsrnradiography has become one of the major techniques of industrial nondestructiverntesting. Although this technique has been well developed,rnimproving both the inspection process and cycle time, it does suffer fromrnseveral drawbacks.rnDigital image processing techniques allow the interpretation of the image tornbe automated, avoiding the presence of human operators making the inspectionrnsystem more reliable, reproducible and faster. Moreover the high-level imagernprocessing methods can even replace the expert's knowledge.rnThis paper describes our attempt to develop and implement neotericrnalgorithms for the purpose of automatic defect detection in radiographicrnimages by digital image processing. The various defects in radiographicrnimages are identified by means of various image-processing algorithmsrnsuitable for defect detection. These are well established algorithms adaptedrnfor use with radiographic images which include those for improving thernquality of the radiographic image, such as reducing image noise andrnincreasing contrast, and algorithms for analysis of image contents, such asrnlocating edges or regions and segmentation from the background image. Thernstandards and type of defects is used to train a neural network. Afterrnpreprocessing, the radiographic image is fed to the neural network. Thernresult is a summary of the type of defect, its extent, size and otherrnnecessary details need for analysis. These defects are thereforernautomatically detected and evaluated with reference to standardrnspecifications.