Tree-plantation decipherment of panchromatic aerial photo images using supervised template matching

Tree plantation decipherment task on panchromatic (black and white) aerial photo images includes both single trees and groves contours recognition. At the first stage of decipherment the locations of these tree-plantations should be defined using template matching. There are, however, inherent problems because of textural template foreground and deformable tree forms. In this paper a new algorithm for supervised textural and template parameter estimation is proposed. It combines the benefits of the pattern analysis and template matching methods. It is described in detail tree-plantation and textural image models. Template matching thresholds for a single tree, a garden bed and a forest are given. An interactive algorithm for iterative mask parameter estimation of templates is proposed.