The quest for contrast in digital images of micro-nano structured polymer blends before and after CO2 foaming: General interest of directional diffusive filtering for edge enhancement

Multiphase materials composed of several phases with smooth interphases and/or sharp interfaces are ubiquitous in polymer science. This study addresses the challenge to identify and segment automatically the right interfaces and interphases in electronic digital images for a sound analysis of the structuration of materials at the micro-nano scale. The selected strategy is to primarily enhance contrast between phases in digital TEM / SEM images for a straightforward subsequent segmentation of phases i.e., by a simple threshold of the grey level intensity histogram. To do so, a nonlinear transient (time dependent) diffusion process is used. In the diffusive process, the grey level intensity spreads non-uniformly across the image pixels i.e., adaptively to local properties of the image. When computed at a large enough time of the diffusive process, the reconstructed image tends to a piecewise constant solution representing a simplified image at a lower resolution with sharper boundaries, bringing the sought local contrast enhancement. The solution image is shown to facilitate the automatic segmentation step and calculation of representative size distributions. This approach is in particular very suited to the characterization of multimodal samples including different scales of porosity. While this (first of two parts) methodological article is useful to multiphase polymeric materials in general, the methodology was here developed for an in-depth study of acrylic block copolymer nanostructures enabling to monitor micro-nano sized foams produced by one step batch scCO 2 green foaming.