Research On Automatic Generation Method Of Groove Defect Image Based On Virtual Reality Technology

As a clean and efficient new energy source,nuclear energy has been widely used in many countries.Nuclear fuel rods are the building blocks of nuclear r eactors.The end-weld welding process of nuclear fuel rods is an indispensable process,because nuclear fuel rods will end in the production process.The occurrence of various defects affects the welding process,so nuclear fuel rod defect detection is nec essary,but the lack of a large number of defective samples in reality has always been a hindrance to the groove defect detection algorithm.This paper studies the automatic generation method of groove defect image,in order to provide more tr aining samples to the groove defect detection algorithm.This paper is funded by the Ministry of Science and Technology's innov ative method special project(project number: 2016IM030300)and the Ministry of Industry and Information Technology's intelligent manufacturin g standardization and new model application(project number: 2016ZXFM02016).The method of automatic generation of groove defect images based on virtual reality technology is researched and explored.The main research works of this work are as follows:(1)Firstly,the causes and classification of the groove defects are introduced.Then,the three-dimensional model of the groove and the digital image processing are used to analyze the mesh features of the three-dimensional model of the groove,the material characteristics and the color features,texture features and shapes of the groove defect images.Feature and spatial relationship features,which are the basis for the generation of identification of groove image based on virtual reality technology.(2)Analyze and build a realistic experimental environment,and use virtual reality technology to build a virtual experimental environment.A refinement groove model was established,and a light source library based on self-illuminating materials was established.The three-dimensional model was deformed using a moving least squares deformation algorithm to create random defects.The 3ds Max software was used to render the groove defect images in a virtual environment.(3)Pre-processing the generated image,extracting the contour of the groove ring contour,using the interpolation method to expand the ring,and u sing the pre-processed simulation image as the sample training Faster R-CNN network to detect the real image to verify the real image.Then,using the simulation image to blindly test the self-template algorithm,the validity and practicability of the proposed automatic generation method of simulation defects are further verified.(4)In order to realize automatic batch generation of defective images,to improve the efficiency of defect image generation,the second development based on MAX Script script language,integrated functions such as model,lighting,deformat ion algorithm and rendering,designed an automatic generation plug-in for editable defect images.The research on the method of groove defect generation not only provides auxiliary effects for this type of defect detection algorithm,b ut also has a reference for other types of defect generation methods.