Urban land-cover classification: an object based perspective

Up to date and accurate urban land cover information is needed in a variety of applications, e.g. urban planning and management. However, depending on traditional surveying tools, especially in large metropolitan cities, to produce such date is a time consuming and expensive task. This has initiated the need to classify remotely sensed data to extract urban land cover information. A new classification approach (object based) has been recently proposed and is currently being investigated. In this research the classification accuracy of object-based classification is tested against statistical classifiers using two images (Landsat and IRS). Results have shown that object based classification yields better classification results.