Facebook Instagram Twitter RSS Feed PodBean Back to top on side

Klasifikácia poškodenia lesa vo veľkej mierke na báze leteckých multispektrálnych snímok a lidarových dát – prípadová štúdia CHKO Dunajské luhy

In: Geografický časopis, vol. 71, no. 1
Tomáš Goga - Hana Bobáľová - Ivan Sačkov - Monika Kopecká
Detaily:
Rok, strany: 2019, 51 - 71
Jazyk: slo
Kľúčové slová:
defoliation, object-based classification, regression model, SVM algorithm, Feature Space Optimization, Dunajské Luhy Protected Landscape Area
Typ dokumentu: časopis/journal
O článku:
Evaluation of forest health conditions is an essential part of forest mapping due to environmental changes in the environment over the last decades. The development of remote sensing technologies creates new possibilities for more effective estimations. The evaluation is performed by estimating the degree of defoliation, and respectively, the depigmentation of assimilation organs of the trees. Based on the relationship between the degree of defoliation as mentioned earlier, and the spectral values, we assume the mathematical relationship between the specified factors. We could also analyse the possibilities of automated methods for defoliation classification. The study deals with methods of calculation of defoliation using 1) a parabolic regression model 2) the Feature Space Optimization tool in eCognition software for object-based image analysis (OBIA) objects and 3) the Support Vector Machine (SVM) classification algorithm for OBIA objects. The research started from the hypothesis that object-based methods of defoliation computation would achieve more accurate and consistent results than a regression model. The study covered the left bank inundation area of the Danube river with an area of 202,38 hectares. The analysis was based on multispectral aerial photos /IR (780 – 880 nm), R (630 – 680 nm), G (520 – 590 nm)/ and airborne lidar data (16 points / m2 total density) obtained on May 2014. Lidar data was converted creating a digital surface model and a raster of intensity using LiDAR Converter Algorithm proceeded to 1x1 m grid structure. The most accurate results were obtained using the SVM algorithm for OBIA objects. The mean error of the absolute loss of assimilation organs is 9.32%.
Ako citovať:
ISO 690:
Goga, T., Bobáľová, H., Sačkov, I., Kopecká, M. 2019. Klasifikácia poškodenia lesa vo veľkej mierke na báze leteckých multispektrálnych snímok a lidarových dát – prípadová štúdia CHKO Dunajské luhy. In Geografický časopis, vol. 71, no.1, pp. 51-71. 0016-7193. DOI: https://doi.org/10.31577/geogrcas.2019.71.1.04

APA:
Goga, T., Bobáľová, H., Sačkov, I., Kopecká, M. (2019). Klasifikácia poškodenia lesa vo veľkej mierke na báze leteckých multispektrálnych snímok a lidarových dát – prípadová štúdia CHKO Dunajské luhy. Geografický časopis, 71(1), 51-71. 0016-7193. DOI: https://doi.org/10.31577/geogrcas.2019.71.1.04
O vydaní:
Vydavateľ: Geografický ústav SAV/Institute of Geography of the Slovak Academy of Sciences
Publikované: 29. 3. 2019