Klasifikace dřevinné skladby z barevného leteckého snímku

Abstract

This diploma work is concerned with the possibilities of analysis of coloured aerial images obtained by traditional methods, as well as by digital scanner and with the possibilities of applying these results in forest praxis.The first part is focused on problems and methods used for identification of tree crowns, description of method of executing picture segmentation and classification of tree species by using the software Definiens Developer. The NIR zone was used by executing the picture taken by digital scanner, second picture taken by traditional way, excluded the NIR zone. In this case the panachromatic spectrum was used. For identification of separate crowns was applied methodology from 2006 [11]. In conclusion of this work are mentioned other data and technologies, which may be used as supporting material in next research of this issue. Wood structure classification run up in Definiens Developer environment, but for examination of clasification correctness was not possible to use data from plane of reference, which would contain the exact position of individual tree species. For identification of separate tree crowns were used pictures resampled to the following resolution - 0,5 m, 0,6 m, 0,7 m, 0,8 m, 1,0 m, 1,2 m, 1.5 m. Both subplots represent 60 years old vegetation. Correctness verification of these results was performed on base comparison of level obtained from automated classification in Idrisi environment and with manually processing vector layer of individual crowns in ArcMap environment.