The study of the network of any issue is important for the economics of each country. The efficacy of the remote sensing itself is directly connected to the capabilities of the applied edge detection method. The idea of this paper is to exploit, view and develop to larger extent the possibilities which are offered by Canny and SUSAN edge detection methods. A short overview about every one of them is shown and discussed. The two chosen methods are compared. In these terms the most effective one is chosen according to the facts connected with the applying of the method. Experimental results as examples are shown to illustrate the advantages and disadvantages of the methods.
The obtained results of the edge detection are classified by Mahalanobis or by Bhattacharyya method, depending of the capabilities of the chosen method in terms to create a self-organizing map which allows the semi-automated system to recognize the object from a remote sensed image. Essential tool in such an aim take under account the self-organizing software system for analysis and conclusions. For real implementation of such system is needed algorithms from different levels, which are discussed, to be chosen. Applying them over the initial image is done by careful and precise varying of the parameters. Future plans are created based on the results.