Image analysis of plagioclase crystals in rock thin sections using grey level homogeneity recognition of discrete areas

Complex optical properties of plagioclase, such as twinning, present a particularly difficult challenge to image processing techniques. Conventional image processing methods attempt to recognize mineral grain boundaries by grey level gradients and are likely to classify optical twin zones as different grains. To overcome this problem, automated grey level processing of small areas under different orientations of cross polarization is used to identify both twinned and un-twinned plagioclase areas as seeds. In an interactive procedure, seeds are projected onto polarized images and are built-up to individual grains. An independent grey level utility supplements automatic processing, allowing grey level profiling down to groups of a few pixels, providing a means for preliminary examinations of thin sections, and to establish typical differences between grain types.