Efficiency of template matching algorithm on GeoEye-1 image for detecting wild pistachio trees and determining their spatial pattern (Case study: The Tag-Ahmad-Shahi protected area, southern Khorasan Province, Iran)

This study was carried out to investigate the ability of template matching algorithm in wild pistachio tree position and their spatial pattern determination using GeoEye-1 images in Tage-Ahmadshahi protected area of Southern Khorasan. After geometric correction and fusion of multi-spectral bands with panchromatic band, three templates were generated for each band and NDVI index. For accuracy assessment of template matching a 100 by 100 meter network (1 ha) used and 60 samples selected randomly and the location of wild pistachio as ground truth registered using a precise DGPS device. For spatial pattern analyzing two useful functions D(r) and g(r) were used. The result of cross correlation showed that the generated templates based on NDVI have the highest correlation between samples The result of template algorithm on satellite image showed that this method with 95.57 percent overall accuracy have the ability to determine tree position by suitable precision on high resolution satellite images. The result of D(r) function showed that wild pistachio trees have nearest neighbor maximum to 45 meter distance. Also the result of g(r) function showed that the trees have a significant uniform dispersion up to 11 meter while between 11.5 and 17 meter have a clumped dispersion which were statistically significant (p