Development of a Method for Forest Type Detection - Softcover

Delineation of forest types was made from 6 scenes of LANDSAT data and validated with National Forest Inventory (NFI) of Germany. Boundary of forest types was used from the official ATKIS vector data to cut out only forest cover. Algorithms for classification were selected to distinguish forest types with training data from NFI and used machine learning approach implemented in caret package of R statistical language. Both pixel based and object based image analysis (PBIA and OBIA) were applied. OBIA resulted the best approach. Mixed behavior was found in the accuracy of the classifications. In general the SVM was the best for 4 of the 6 scenes under evaluation. KNN and RF resulted the best for the rest of the scenes. General schema of the procedure is presented and tips for using every classification algorithm are disused.

Juan Ygnacio López Hernández born in Carora in 1967, Lara State, Venezuela. His degrees are: Forest Engenier (1993) University of Los Andes (ULA) Mérida, Venezuela, MSc ULA (1996) and PhD at Albert Ludwigs Freiburg University, Germany (2012). Today is head professor at ULA and chair of the Photogrammetry and Remote Sensing laboratory.