Grain identification of polarising images with level set method

Abstract

We extend the classical active contour to segment two optical polarizing microscope images of a thin section within the same view. The polarizing microscope is a basic tool employed by many geologists to identify minerals, estimate grain size or shape and to obtain modal percentages of minerals in thin sections. Two polarising images are captured with different light phenomenon: under plane-polarised light, many mineral grains appear as colorless; under cross-polarised light, mineral grains show varying colors and intensities. Automatic detection of mineral grains within images captured polarized microscope is a difficult task. Instead of segmenting any single image with classical active contour model, a novel energy functional is constructed for locating edges of grains with two polarising images. The corresponding Euler-lagrange Equation is obtained with level set method. The evolving curve stops on the edges of the identified grains. Experimental results of the real thin section images are shown.