An Automated Microstructural Analysis of SEM Images for Multiple Domain Sets Using Improved CURE (i-CURE) Algorithm in Machine Learning

In real time materials plays a vital role in all fields. Scanning Electron Microscopic (SEM) images are used for imaging the materials and it provides microstructural information. Scientifically the material strength is analyzed only by microstructural and it helps to understand the applicability range. The demand of good and quality material is increasing day by day. Mostly microscopic images give the result as a qualitative analysis of its material which is not in the form of quantitative. Currently most of the reported work says SEM image analysis was done physically so it takes so many days to complete the process and the result is qualitative. So the material need quantitative analysis which is done only by microstructure. The SEM microstructural analysis helps to predict the materials behavior and strength. Materials behavior cannot be understood without characterizing microstructure because of its heterogeneity characteristics. In proposed work developed an automated tool with CURE (Clustering Using Representatives) algorithm in machine learning is to represent statistical analysis of microstructure predict the performance of the materials. There are different microstructural features in terms of size, shapes, position, or orientation. The proposed automated microstructural image analysis methods give the result in statistical/quantitative so that user/experts can predict the result accurately. The different statistical analysis was analyzed like black particle extraction, histogram, scatterplot, 1D gaussian distribution. The quantitative analysis has made a great solution in finding material or tensile strength of material behavior.