Kauzhiro MORI

Study of Method of Recognizing Trees Automatically in Digital Aerial Photograph

Atsushi RIKIMARU, Yukio MUKAI

Global warming caused by deforestation increasing that number of the greenhouse effect gases has brought about serious effects to global environment, therefore we should immediately protect forests absorbing the greenhouse effect gases. The purpose of this study is to develop a method of recognizing trees automatically using remote sensing data for measuring the amount of the forest resources. And the position and the number of trees in a digital aerial photograph were presumed by using the template matching. The results of this research are as follows.

1) The result by using the template matching with a template generated by the image cut out from a digital aerial photograph shows a high correlation in the places where the trees exist. Therefore, the template matching is effective as a method of recognizing trees.
2) The result by using the template matching with a template generated by averaging the tree images in five places cut out from a digital aerial photograph shows a higher correlation than that of the case with a template cut out from one place. So averaging the tree images cut out from the plural places can generate an objective template.
3) The result by using the template matching with the method of cutting out the tree images for averaging automatically shows the same correlation as the result with a template generated by averaging beforehand. The positions of the trees were presumed based on this, and accuracy of the presumption computed using the positions of the trees recognized by visual interpretation on the digital aerial photograph were 71.0% or less.
4) The result by using the template matching with a template generated from a tree model by the computer graphics aiming to generate an objective template is show a high correlation in the place where the trees exist. Therefore, the tree model is effective to the template matching.