SEM Microscopic Image Segmentation Based on Markov Field Models

The Scanning Electronic Microscopy (SEM) images are usually used to analyze a certain kind of material properties. To get automatic and accurate quantitative research, image processing methods are utilized to analyze surface morphology from the images obtained from SEM. In this paper, we focus on the Activated Carbon Fiber (ACF) SEM material images. K-nearest neighbor smoothing and Lapalacian sharpenning methods are performed for preprocessing. This paper presents a segmentation method based on Markov Field Models algorithm to separate the pores and background for the material image. Iterated Conditional Mode (ICM) priority iteration algorithm is applied to find exact optimal estimators for labeled field and the model parameters. The experimental results show that the segmentation algorithm based on Markov Field Models is appropriate for microscope images.