Classifying Land-Cover Using Texture Statistics

We present a method for labeling the land-cover in an aerial image using image texture statistics and a trained neural network to interpret the texture data. The introduction includes a brief discussion of land-cover mapping and the types of land-cover classes we would like to identify. There is a section describing the texture statistics used and the Gray Level Cooccurrence Matrix (GLCM) for calculating them. There is also a section describing neural networks and the specific parameters used for our neural networks. The results section contains the images used for training the neural network, example images for the texture statistics, and labeled aerial images.