Ещё раз об автоматической расшифровке рентгеновских снимков сварных швов

Abstract

Introduction. This article is dedicated to a possible approach to automatic inspection of welded joints X-Ray images. This problem is very urgent but there is no established practice of the automated X-Ray images inspection and image objects recognition. The purpose of the article was to highlight the tasks included in the X-Ray interpretation process that can be automated, and to offer possible solutions. 
 Method. Analysis and interpretation of images were made using "X-Vizor" software by “Newcom-NDT”.
 Results. Studies have shown that the straight and elliptical seam areas could be recognized automatically. It provides the possibility for a signal-noise ratio to be automatically measured. Wire image quality indicators and grooving gages also could be recognized automatically in approximately 80% cases. So it is possible to automatically calculate the sensitivity of the inspection. It is also achievable to automatically determine the basic spatial resolution using duplex wire image quality indicator. But the task of automation of defect recognition is still not solved. The possible alternative was offered. It is semi automation of defects recognition and analysis. This procedure needs the human to take part in image interpretation process. 
 Conclusion. The conclusion is that the total automation of digital X-Ray image interpretation of welded joints is now unavailable. Nevertheless using automation algorithms in all parts of interpretation process where it is possible can highly increase the performance of the inspection and its reliability.