Research On Image Segmentation Technology Applied In Titanium Alloy Metallographic Quantitative Analysis

Since the 1980s, the titanium alloy, due to its excellent performance, has been begun more and more widely used in various fields of the industrial development of modern society. There are some math relationships between mechanical properties and microstructure of titanium alloy. By measurement of the various parameters, processing and analysis of the metallurgical microstructure, microstructure analysis technology can achieve the goal of quantitative analysis to the relevant metallurgical characteristics.Metallographic analysis is a discipline of studying the microstructure of metallic materials and finding the relationship between microstructure and its macroscopic performance. It can promote the metallographic analysis to be automatic and quantitative by means of digital image processing techniques to help metallographic analysis. Image segmentation is the key step in metallographic analysis based on image processing technology, but there is no general image segmentation algorithm for all metallographic images due to the different features of every kind of metallographic image.Therefore, in this thesis, according to the characteristics of the titanium alloy metallographic image, the traditional edge segmentation and threshold segmentation algorithm are firstly studied, then some segmentation algorithms based on the linear fringe direction of titanium alloy metallographic texture and binary morphology are proposed, and implements effective segmentation of the titanium alloy metallographic image. Finally, several mathematical algorithm models of titanium alloy metallurgical microstructure characteristic parameters measurement for titanium alloy metallographicthe quantitative analysis are initially established. These researches have laid a good foundation for the quantitative metallographic analysis of titanium alloy. The following is the main results of this thesis:Firstly, the image of the titanium alloy metallurgical are preliminary studied which abtained the relationship between the texture of titanium alloy metallurgical characteristics and internal microstructure features. Based on that, the suitable preprocessing algorithm of titanium alloy metallographic image are determined.Secondly, segmentation algorithm to titanium alloy metallographic image is studied. It is shown that the results obtained by using edge segmentation algorithm and threshold segmentation algorithm to typical low magnification titanium alloy to metallographic are not very good. Then, through the study of low magnification image texture features of titanium alloy microstructure, the segmentation algorithms based on the frequency domain direction and GLCM are proposed, and implements a more satisfactory segmentation result.Thirdly, image segmentation methods based on binary morphology for the titanium alloy microscope metallographic image are studied. Binary morphology is applied to edge detection and skeleton treatment for the initial segmentation results of titanium alloy metallographic image. to remove the noises in the segmentation image effectively, the structure elements are separately designed according to the features of the isolate point, endpoint and bifurcate point generated in the segmentation image, and the image segmentation result is improved greatly.Then, based on the segmentation result, several mathematical algorithm models of titanium alloy metallurgical microstructure characteristic parameters measurement for titanium alloy metallographicthe quantitative analysis are initially established.Finally, the proposed algorithms, some common image segmentation algorithms and preprocessing algorithms are developed in Matlab and Visual C++environment, based on OpenCV, and all these lay a foundation for the further quantitative analysis system for titanium alloy metallographic.