Evaluation of using Russian Cosmos satellite image for land-use in civilian regions

ABSTRACT

This research represents and attempt to evaluate the ability of using Cosmos satellite image for land-use classification in one of the most populated region in Baghdad [Al-Sheikh'Omar], which characterized by irregular land-use and different environments. Such variety can be recognized by wide range of image brightness values and interfering between classes [asphalt, metal roofed galleries ' etc]. this small area represent different types of land-use [industrial, commercial, governmental, and housing], and that point to exist of different subclasses in the main classes. Erdas program used to treat Cosmos image cover the pointed region, which include digital processing and analysis firstly, and digital classification secondly. The first process includes making radiometrical enhancement [contrast and density slicing], so as to separate high buildings fro metal roofed galleries, in addition to spatial enhancement [edge enhancement and spatial filtering] to get out back street of [1.8] m wide or more further to main street. While digital classification includes applicating of three types of classification [supervised, unsupervised, and hybrid]. The study concludes that the hybrid classification has a good ability to avoid the difficulty of training all land-use categories in one hand, and to treat the interference between them on the other. So hybrid classification is important to separate street from houses, buildings and metal roofed galleries, and consider as the best between others. While the unsupervised classification was more simple, clear and object, that make use of it in determining the identifying regions for the supervised classification which in turn very useful and accurate in getting out houses