Research On Information Extraction Algorithm Of Alloy Microscopic Image Based On Image Analysis

The microstructure distribution of alloy determines its properties.For the aim of regulating the alloy microstructure to improve its properties,it is essential to extract and analyze the microscopic image information of alloy.Facing the problem that there are great differences in different alloy microstructure images caused by the diversity of alloy composition,the existing alloy microstructure image information extraction algorithms cannot meet the information extraction requirements of tellurium copper alloy microstructure image and cupronickel alloy microstructure image.This paper focuses on how to deeply extract the microstructure image information of tellurium copper alloy and cupronickel alloy.For the alloy microstructure images of different structures,two microstructure image information extraction algorithms are proposed,and the cupronickel alloy microstructure images with line information are deeply studied.The relevance between the image information and the alloy properties is interpreted by establishing the corrosion resistance prediction model.The principal research are as follows:(1)Information extraction algorithm of tellurium copper alloy microstructure image based on simplified Deep Labv3+.In order to study the microstructure in the tellurium copper alloy,tellurium phase segmentation model and the extraction algorithm of tellurium phase area and position based on connected region are proposed.In the tellurium phase area extraction algorithm,the tellurium phase area is calculated by labeling the tellurium phase area and counting the labeling results.In the tellurium phase position extraction algorithm,the centroid coordinates of the tellurium phase region are used as the tellurium phase position,and the tellurium phase position is obtained by calculating the centroid of the marked transverse and longitudinal coordinates of the tellurium phase region.The experimental data is the microstructure image of tellurium copper alloy,and the experimental results of image information extraction show that the improved model has shorter consumption time,better performance and better prediction results.The accuracy of tellurium phase area extraction algorithm is 92.8%.The proposed algorithm can solve the problem that the tellurium phase area and position in the microstructure image of tellurium copper alloy cannot be calculated.(2)Information extraction algorithm of cupronickel alloy microstructure image based on convolution operation.In order to study the microstructure of cupronickel alloy,trigeminal angle recognition and angle extraction algorithm are proposed.In the trigeminal angle recognition algorithm,16 different convolution kernels are defined,the images are convoluted respectively,and the convolution results are judged pixel by pixel to recognize the trigeminal angle.In the trigeminal angle extraction algorithm,by recording the trigeminal angle branch coordinates starting from the intersection of trigeminal angles,the branches of trigeminal angles are fitted by linear regression,points are taken on the fitted straight line to construct triangles,and the trigeminal angle is calculated according to the cosine theorem.In order to verify the effectiveness of the microstructure image information extraction algorithm of cupronickel alloy in practical application,the corrosion resistance prediction model based on trigeminal angle recognition and angle extraction algorithm is proposed.The maximum connected length,the number of trigeminal angles and the angle distribution of trigeminal angles in the cupronickel alloy microstructure image is calculated by the connected area marking method,trigeminal angle recognition algorithms and angle extraction algorithms separately and the corrosion resistance prediction model of the cupronickel alloy is established by giving different weights.The experimental data is the microstructure image of cupronickel alloy,and the experimental results of image information extraction show that the accuracy of trigeminal angle recognition and angle extraction algorithms are 100% and 96.9% respectively.The proposed algorithm can solve the problems that trigeminal angle cannot be recognized and angles cannot be calculated in the microstructure image of cupronickel alloy.The accuracy of the corrosion resistance prediction model is 81.88%.The trigeminal angle recognition and angle extraction algorithm can provide an effective means in the study of microstructure and properties of cupronickel alloy.The alloy microstructure image information extraction algorithm based on image analysis solves the problem that the image information cannot be extracted in the alloy microstructure study,especially for the microstructure images of tellurium copper alloy and cupronickel alloy or other alloy microstructure images with the same image information.At the same time,the algorithm research establishes the relevance between the alloy microstructure image and the properties,which provides effective means for regulating the alloy microstructure and improving the properties of the alloy.The research on the microstructure image information extraction of tellurium copper alloy and cupronickel alloy effectively verifies that computer vision technology can play a good role in the field of alloy microstructure image.Due to the particularity of alloy microstructure image information,the alloy microstructure image information extraction algorithm also enriches the development of computer vision.