Research And Development Of On-line Image Detection System For Two-sided Crack Propagation Of Fatigue Samples

Accurate measurement of crack length is very important in fatigue crack growth rate test.The traditional visual inspection method requires manual shutdown detection,which consumes labor and has subjective measurement errors.However,most of the existing image-based fatigue specimen crack detection systems are only suitable for single fatigue specimens in the relevant national standards and are not universal.Moreover,more systems use a single camera for single-side detection,which cannot effectively judge the symmetry and effectiveness of the crack propagation on both sides of the specimen before and after,and has limited practical application value.Therefore,in this paper,six types of fatigue specimens specified in GB/T 6398-2017 "Fatigue Crack Growth Method for Fatigue Testing of Metallic Materials" are taken as research objects.Considering the symmetry and effectiveness of crack growth,a hardware testing platform is built using two Dalsa industrial cameras and related accessories.Based on Visual Studio 2015(VS2015)software development platform,online measurement of crack length value and calculation of crack growth rate are realized.Through the test verification,the system has reached the testing precision requirements specified by the national standard.The main contents of this paper are as follows:(1)According to the system requirements,the software and hardware platform design scheme of the online image detection system is determined.Use camera,lens,light source,GIE62+acquisition card,industrial computer and other equipment to build a hardware platform.With C++ as the programming language and OpenCV image tool,the main function design and development of the system software are completed.(2)In camera calibration,Zhang Zhengyou method is used to calibrate the camera,correct camera distortion and obtain length conversion factor K;According to the need of automatically measuring crack length,an automatic region of interest extraction and Otsu threshold method is proposed.By comparing the traditional filtering method with the adaptive median filtering method,the median filtering method is improved.(3)The contour search method is used to automatically select the initial seed points in the crack area identification,thus avoiding complicated operations;The region filling connection algorithm and the jump connection algorithm are respectively used for special cracks that are opened and broken,thus improving the accuracy rate of crack region identification.Sub-pixel subdivision technology is used to realize superpixel detection,which further improves the detection accuracy.The zhang parallel thinning algorithm is improved to extract the crack skeleton with single pixel width under the condition that the crack is not shortened,thus reducing the measurement difficulty of the crack length value.(4)In the fatigue crack growth rate test of CT sample,CCT sample and SENT sample,the measurement accuracy,reliability and adaptability of the detection system developed in this paper are verified by comparing with COD gauge measurement value and crack pixel length value.