|
Parallel with the development of remote sensing and high resolution satellite images major challenge for modern research has become almost to automate the classification of the data obtained. One of the most rapidly developing methods for automatic classification is object-oriented image analysis (OBIA, Object Based Image Analysis). The aim of the present study was to use the OBIA method to create the current land cover map which is part of the documentation necessary for new water power-station on the middle part of Vistula river. In this paper the RapidEye satellite images (5 spectral bands, two in the NIR range) covering an area of about 5 300 km2 and eCognition Developer (TRIMBLE) software were used. As a result of the analysis and land cover map was obtained, represented by 29 classes. The largest area is covered by agricultural land (59.5%; arable land – 35.52%) and forests (29.1%; mainly coniferous 21.4%), reflecting the rural – forestry character of the area. Analysis of the accuracy of the obtained results has shown that the OBIA method gives quite good results (Kappa coefficient equal to 0.8) for land cover mapping of central part of Poland based on the RapidEye imageries.
|