Detection of Korean Pine and Extraction of Korea Pine Space Coordinates Using Large-Scale Aerial Photographs

Abstract

For estimation of single tree parameters using 1:6000 large-scale aerial photographs, tree species identification is an important starting point. This paper presents a new approach for identifying tree species, delineating individual trees and extracting single tree space coordinates in coniferous and deciduous forests of Liangshui National Nature Reserve of P. koraiensis, Northeast of China. To identify tree species and isolate Korean pines, the extended knowledge classification method was applied with several different auxiliary variables. For delineating individual Korean pines, a vector to raster algorithm was applied and a crown vector polygon layer was generated. The bar centric coordinates of the crown vector polygons were extracted as Korean pine space coordinates. Thereafter, the estimated data were compared to the field measured data and the interpretation accuracy had three categories with different diameters at breast-height. In our study, the best classification accuracy of P. koraiensis was 94% with knowledge classification, increasing by 9% over that of supervised classification. P.koraiensis had the best interpretation accuracies of 10.2%, DBH ranged from 6cm to 18cm; 48.9%, DBH from 20cm to 30cm; and 90.4%, DBH more than 30cm, respectively.