Research On Automatic Detection Andrecognition System Of Weld Defects Inx-ray DR Image

Digital X-ray imaging is the mainstream of the future radiography testing,the traditional manual evaluation method is not suitable for evaluation of digital image.With the development of computer technology,automatic detection and recognition is a hot issue of weld defects in nondestructive testing.DR image of a large size girth weld was considered as research object,the automatic detection and recognition of weld defects was carried out,the automatic detection and recognition software system of DR images in weld defects was developed based on Visual Studio 2010 and opencv3.0 open source code.The main research works is listed below:(1)the contrast of DR image was improved by preprocessing algorithm,and a new algorithm for automatic detection of weld defect in DR images was proposed.Firstly,the smooth radius r was set,the smoothing template of(2r+1)×(2r+1)was constructed.Then the simulated ideal weld image was created by image median filtering.and image subtraction operation was performed between ideal weld image and original image.Finally,look for all the areas where the subtraction difference exceeds gray connectivity(the given threshold)as a suspect defect.The influence of gray connectivity and smoothing radius on the automatic detection of weld defects was analyzed.(2)The characteristic parameters of all suspicious defects were analyzed and calculated based on the binary image generated by the automatic detection of weld defects and the original grayscale images,and nine characteristic parameters and corresponding algorithms were proposed,and the corresponding results were obtained.The qualitative analysis algorithm of weld defects was proposed,according to the defect characteristic parameters,and the automatic qualitative analysis of weld defects were realized.(3)The design and integration of the automatic recognition software system of weld defects in DR image was performed.Firstly,the overall framework design of automatic recognition software system of weld defects was carried out.Secondly,the main interface of the automatic recognition system,the image geometrictransformation module,the image preprocessing module,the defect detection module,the defect statistics and the quality rating module were designed and realized.Finally,a set of software system for automatic recognition of weld defects in X-ray DR images was completed mainly by Visual Studio 2010 development environment,the main functions of software system including: import of DR acquisition images,geometric transform of DR images,preprocessing of DR images,automatic detection of weld defects,parameters calculation of weld defects,qualitative analysis of weld defects,display list of defects,real-time location of defect,feature editing of defect,defect statistics,description of special defect,display of exceed defect information,automatic quality rating,information storage and so on.(4)Typical weld defect images,such as dispersed porosity,dense porosity,tungsten inclusions,inadequate penetration and so on were used to test for software system,the software test results show that the automatic detection results of weld defects are consistent with the actual defects when the detection parameters are reasonable,the defect qualitative analysis algorithm is reasonable and the software runs well.