Application of a\nHessian-Based Image-Processing Method\nfor Enhanced Visualization of Nanoscale Rubber Cross-Linked Network\nStructures from Electron Microscopy Images

Characterizing rubber cross-linking, nanoscale structures\nthat\nsignificantly influence rubber’s mechanical properties, is\ncrucial for advancing the design of rubber materials, and electron\nmicroscopy is essential for observing the cross-linked networks. However,\nreliance on manual markup to make the network visible introduces subjectivity\nand hinders the objective analysis of many samples. This study presents\nan image-processing method based on the Hessian matrix to enhance\nand extract cross-linking network structures. We confirmed that this\nmethod effectively enhanced network structures across a wide range\nof cross-linking levels. From parameters extracted from enhanced images,\nwe derived an equation to predict the cross-link density and mechanical\nproperties. This equation outperformed the use of image luminance\nalone, thus confirming the effectiveness of the image enhancement\nmethod in characterizing cross-linking. We successfully estimated\nnanoscale cross-linked network structure parameters directly from\nimages previously measured indirectly. This versatile approach, applicable\nto various structures, is expected to significantly improve nanoscale\nstructural analysis, thereby paving the way for the development of\nhigher-performance materials.