Fast calculation of co-occurrence matrix parameters for image segmentation

Abstract

Image segmentation methods based only on grey level information are not suitable for pictures in which regions exhibit almost the same average grey level and differ only for local variations or texture. By extending these methods to textural features, better results are expected. Among textural features, those extracted from co-occurrence matrices are quite effective. A fast algorithm for the calculation of these parameters for windows centred on each pixel of the image is presented.