Improving the Efficiency of Defect Image Identification During Computer Decoding of Digital Radiographic Images of Welded Joints of Hazardous Production Facilities

This article is devoted to improving the efficiency of flaw image identification during computer decoding of digital radiographic images. The paper studies the problem of segmentation of flaw images. Models of segmentation of flaw images on a radiographic image are studied for both manual and computer decoding. The difference between algorithms for searching and identifying groups, clusters, chains of pores, slag and metal inclusions from manual decoding of images is shown. Algorithms for the search and identification of flaws for use in digital radiography complexes have been developed and experimentally tested at HSC KARS. The convergence of the results of computer and manual decryption was 0,85.