Boron Nanoparticle Image Analysis using Machine Learning Algorithms

The effort of digital image processing involves efficient computation aimed at developing an economical, faster and more accurate, and cost-effective automated system. The objective of this paper is to ascertain and categorize Boron nanoparticles (BNP) using digital image processing techniques. The spatial features are unsheathed from the Boron nanoparticle Transmission Electron Microscope (TEM) images using different segmentation techniques, namely; Fuzzy C -means (FCM) and K-means. The size of Boron nanoparticles is determined and categorized based on the area(size) in the microsize.The synthesization and characterization of Boron nanoparticles play an important role as an elementary procedure for the formation of Boron nanoparticles. The results are analyzed, interpreted and comparison is done with the manual values to observe the efficacy of the results. It is observed that the K-means segmentation technique yields a smaller amount of error (5.87%) as compared with Fuzzy C-mean(16.78%). Hence, it is considered that the K-Means is the most relevant segmentation technique for Boron nanoparticle image analysis and categorization. The statistical test of significance is applied using the Chi-square testing method (at 5% of significance level) to check the relationship between the manual results and the algorithm results.The proposed study also establishes collaborative research work between Chemistry and Computer Science departments to develop computational research on these platforms.

Automated summary

The objective of this study is to determine and categorize Boron nanoparticles using digital image processing techniques. Different segmentation techniques are employed to extract spatial features from Transmission Electron Microscope images, and the nanoparticles are categorized based on their size. The results show that the K-means segmentation technique is more effective for Boron nanoparticle image analysis and classification.