Analysis of remotely sensed data by means of logic filters

Multi-disciplinarily, various scientific research fields need multi-temporal information about land use/cover rates. Especially in metropolitan areas quick data acquisition and processing is important due to the dynamic improvement. Nowadays, remotely sensed image data have considerable use in classifying the earth surface for land use/cover. Recent studies involve temporal detections, analysis and determinations of urban morphology using satellite sensor images. The aims of this paper i) to develop an algorithm using Matlab&Simulink program in order to filter Vegetation, Impervious and Soil (V-I-S) components ii) to analyze land use/cover by means of Landsat 5 TM image of region in Istanbul by using developed algorithms. Selected area from Landsat 5 TM image filters vegetation, impervious, soil and water components according to upper and lower Threshold values obtained under user supervision based on each bands histogram. If that component exists on the image then "1" and else if not "0" is recorded for the pixel on the new image. This procedure is repeated for each band. The results obtained using the suggested algorithm compared with the supervised classification results using Erdas Imagine Program. Algorithm is tested and proved in application for medium resolution satellite sensor images.