Challenging Micrograph Analysis: The Latest in Computer Vision

Micrograph analysis of advanced materials is a challenging problem due to ever-changing detection and measurement requirements. It is nearly impossible to provide a single 'black box' solution to solve all analysis needs, even within a single material system. MIPAR's approach is to provide customized, robust, and problem-specific solutions that overcome real challenges in automating micrograph quantification, that are uniquely tailored to each case. In this webinar, we will demonstrate how some of the most challenging problems in micrograph analysis are solved using MIPAR. 

• Twinned grain analysis is a notorious problem in materials science due to almost identical contrast between twins and grains. Using next-generation machine learning techniques, MIPAR was able to reliably detect the twinned grain structure and measure its characteristic size
• Another analysis problem, quite popular in additive manufacturing and powder metallurgy, is shape characterization of loose powder particles. Using MIPAR’s powerful algorithm development environment, a solution to these challenges was devised, leading to improved confidence in quantifiable powder batch differences that are often difficult to tease out.
• Lastly, nanofiber analysis is popular in many manufacturing applications, where fiber thickness, length and porosity strongly affect material properties. State-of-the-art deep learning technology allowed for fiber/background segmentation despite similar gray values, and MIPAR allows the user to generate thickness distributions, intercepts measurements and porosity densities in the entire network, leveraging statistical power to solve this challenging problem.