Structure-Preserving Segmentation of Individual Tree Crowns by Brownian Motion

Abstract

A fast, simple and easy to implement segmentation method is presented. A particle is walking in the image in the fashion of a Brownian motion to create the segments. Using this method, it is possible to segment the image and find the inner structure of the segment at the same time. The inner structure could be useful for further analysis. Results on how the method performs is shown by segmentation of tree crowns in aerial images. Although the method is presented in two dimensions it is easily generalized to three dimensions.