Terrain classification using multiple image features

A wide variety of image processing applications require segmentation and classification of images. The problem becomes complex when the images are obtained in an uncontrolled environment with a non-uniform illumination. The selection of suitable features is a critical part of an image segmentation and classification process, where the basic objective is to identify the image regions that are homogeneous but dissimilar to all spatially adjacent regions. This paper proposes an automatic method for the classification of a terrain using image features such as intensity, texture, and edge. The textural features are calculated using statistics of geometrical attributes of connected regions in a sequence of binary images obtained from a texture image. A pixel-wise image segmentation scheme using a multi-resolution pyramid is used to correct the segmentation process so as to get homogeneous image regions. Localisation of texture boundaries is done using a refined-edge map obtained by convolution, thinning, thresholding, and linking. The individual regions are classified using a database generated from the features extracted from known samples of the actual terrain. The algorithm is used to classify airborne images of a terrain obtained from the sensor mounted on an aerial reconnaissance platform and the results are presented.