A comparative study of linear and nonlinear edge finding techniques for Landsat multispectral data

Several numerical methods were computerize in order to evaluate edge finding techniques for accuracy, versatility and computer time and cost. The study was preformed primarily to select edge detection operators which will be used to do spatial correlation between images for image registration of Landsat Multispectral Scanner (MSS) scenes. These method are also used to enhance and discriminate certain types of boundaries; such boundaries or edges are related to geologic structure, topographic relief, albedo and color changes. Application of these techniques has an aid in geologic and geographic remote sensing is evaluated. Landsat scenes of Washington, D.C. and Death Valley were chosen to test the applicability of techniques both in arid desert and vegetation--water--urban cases.