Exploring the effectiveness of various texture features for change detection in remote sensing images

Change detection process can be considered as a classification problem, classifying the image into change and no-change classes. Change detection in remote sensing images is an active research area in applications such as disaster management, urban planning etc. Texture is an important characteristic which can be used to identify various objects or regions in an image. Texture features along with spectral and morphological operators have been shown to have a positive impact on the accuracy of change detection in remote sensing images. Haralick, 1973 defined a set of twenty eight texture features which can be extracted from any image on a block basis. This paper explores the effectiveness of various texture features in enhancing the accuracy of change detection process. In this paper we have used three sets of texture features and their impact has been analyzed, keeping all other parameters constant.