Defect detection method, storage medium and device for radiographic image weld area

Abstract

The invention discloses a method, storage medium and equipment for detecting defects in a radiographic weld seam region. A residual map is generated based on background characteristics, and suspected defect regions are detected on the residual map, and the algorithm is avoided when the algorithm is sensitive to noise and background grayscale. At the same time, a high defect detection rate is guaranteed; the multi-scale candidate frame and convolutional neural network classification are used to classify the suspected defect areas detected by the residual map, which can solve the problems of complex and changeable defect shapes and outlines and the difficulty of effective feature extraction, and obtain a lower rate. False alarm rate; the use of non-maximum value suppression for classification results can ensure accurate positioning of suspected defect areas, facilitate further defect identification and defect size quantification, and at the same time give the confidence of the final detection results, providing reviewers with more full reference information.

Description

technical field [0001] The invention belongs to the technical field of weld defect detection, and in particular relates to a method, storage medium and equipment for detecting the presence or absence of defects in a radiographic weld region based on background characteristics. Background technique [0002] With the continuous development of welding technology, the welding quality has been greatly improved. Generally speaking, the proportion of defective radiographs in the total number of negatives is less than 5%. If it is possible to automatically and accurately perform defect detection based on the weld area before film review, and select all radiographic weld images (5%) with suspected defects, then the subsequent film review personnel only need to perform this small part of radiographic weld images By performing defect classification and evaluation, the efficiency of film review can be further improved, and the work intensity of film review personnel can also be greatly ...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide a method, storage medium and equipment for detecting whether there is a defect in the weld area of ​​a radiographic image based on the background characteristics, and to generate a residual map based on the background characteristics. The detection of suspected defect areas on the difference map avoids the sensitivity of the algorithm to noise and background grayscale while ensuring a high defect detection rate; the suspected defect areas detected in the residual map are classified based on multi-scale candidate frames and convolutional neural networks. It can solve the problem of complex and changeable defect shape and outline and the difficulty of effective feature extraction, and obtain a lower false alarm rate; use non-maximum value suppression for classification results to ensure accurate positioning of suspected defect areas, and facilitate further defect identification, Defect size quantification and other operations, while giving the confidence of the final detection results, providing more sufficient reference information for film reviewers

Images