Optimization based porosity segmentation

A porosity segmentation technique has been proposed in this work. Scanning Electron Microscopy (SEM) is often used as a non-destructive technique to obtain microscale images of rock samples. These images pose various difficulties before the user. Unlike sandstones, carbonates display a large heterogeneity in their pore-size distribution owing to their evolution. Additionally the quality of the image often mandates several image processing steps before they can be used for interpretation tasks. This variation appears due to irregular distribution of intensities in the pixel values of the SEM image. Conventional techniques of global binarization used in extracting the porous part suffer due to this reason. This work introduces a novel method of optimizing the various image processing parameters for proper extraction of porous part. The proposed method implements the Simulated-Annealing (SA) based global optimizer for finding optimum values of these parameters. Results of the application on five SEM images have been shown. The images are processed with optimum choice of parameters after preprocessing. Finally, we report the total porosity in terms of pore and throat porosity.