Implementation of semi-automated object-based image land cover classification methods: a case study of the Male Karpaty Mts. (Slovakia)

Year, pages: 2017, 113 - 128

The semi-automated object-based image classification (OBIA) has recently become a key method in land cover mapping. Our study demonstrates this robust procedure using OBIA analysis obtained by LANDSAT 8. The advantages of this classification system include time-efficiency and images of greater detail with the same quality as a manual interpretation. A delineation algorithm was created in the Malé Karpaty Mts. with 93.6% overall accuracy and 0.81 Kappa coefficient. Our process was based on multiresolution segmentation that was created on 4 hierarchical levels, with emphasis placed on the expected segmentation result and classification with the following applied spectral indices: the normalized differenced vegetation index, simple ration index and built-up area extraction index. We delineated 11 dominant land cover classes of the Malé Karpaty Mts. on the 4th hierarchical level of our classification, with emphasis on forest and semi-natural areas. This procedure was concurrently applied to other Western Carpathian ranges with the best overall accuracy determined at 78.4% in the Slanské vrchy Mts., 74.16% in the Tríbeč Mts., and 73.1% in the Považský Inovec Mts.

Gábor, M., Karlík, L., Falťan, V., Petrovič, F. 2017. Implementation of semi-automated object-based image land cover classification methods: a case study of the Male Karpaty Mts. (Slovakia). In Geografický časopis, vol. 69, no.2, pp. 113-128. 0016-7193.

Gábor, M., Karlík, L., Falťan, V., Petrovič, F. (2017). Implementation of semi-automated object-based image land cover classification methods: a case study of the Male Karpaty Mts. (Slovakia). Geografický časopis, 69(2), 113-128. 0016-7193.