Research And Application Of The Algorithm Based On Closed Solution For Metallographic Quantitative Analysis

Matching technology originates from the demand of photo processing and film shooting.How to extract foreground objects from digital video or static digital images is a key research topic in the field of image and video editing in recent decades.The technology of accurately extracting foreground objects in the scene is called digital matting.With the rapid development of computer hardware and software in recent years,digital matting method has been widely studied.Because there is no environmental restriction on the natural image matting method,it can be applied to any natural image,and has a wide range of applications,but this kind of method is often complex and difficult to guarantee the accuracy,so it is still a hot spot in the digital image processing neighborhood.In the field of metallographic quantitative analysis,the basic image processing of the collected metallographic images is carried out.Supported by the basic principle of stereology,the threshold segmentation method is usually used to quantitatively analyze and study some characteristics of the microstructure in the images,such as area,quantity and shape,so as to give the materials.Accurate evaluation of application performance of components or parts.Based on my previous work experience,this paper studies and analyzes the natural image matting technology,and applies matting technology to metallographic quantitative analysis to further improve the accuracy of quantitative analysis results.The research work in this paper is as follows:1.Studying the "extra information matting" technology,the basic principle and related methods of natural image matting,realizing the classical matting algorithm of closed-loop matting and experimenting the effect of using natural images such as characters and hair.2.Closed solution matting algorithm,kernel function matting algorithm and iterative boundary search matting method are studied.Aiming at the problem that the precision of the closed-solution matting algorithm is not high in the actual matting,this paper proposes to use the joint kernel function and iterative boundary algorithm to pre-process the basic image matting.Through the pre-processing,most of the outline information in the foreground is obtained.Then,in the case of the poor effect of the transparent object matting,Combining with the closed-loop matting method for further matting processing,this method is called matting algorithm based on closed-loop solution.3.The metallographic analysis system is developed by using Delphi and MATLAB OLE mixed programming.The core algorithm is matting algorithm based on closed solution.It is used to solve the problem of picking up the background in the transparent area and improving the accuracy of quantitative analysis results.Software functions include: grayscale and binary images,corrosion and expansion algorithm,open and close operations,refinement of grain boundaries and grain boundaries extraction,fixed threshold segmentation algorithm and closed-loop matting algorithm.4.Data experiments are carried out by the metallographic quantitative analysis software developed in this paper: 1)Experiments are carried out on different natural images,and better results are achieved by adjusting the parameters.2)Through the quantitative metallographic analysis of the pearlite phase composition ratio and the porosity of copper-based porous materials,the difference between the closed-loop matting algorithm and the fixed threshold segmentation algorithm is experimentally studied.