Young oil palm recognition using Haar features based on worldview-2 images

Abstract

This paper presents application of Haar-based rectangular features for recognition of young oil palm tree based on WorldView-2 imagery data. Younger age of oil palm tree should be more separated and sparse in the imagery data compare to the matured age of oil palm tree. Even though that the young oil palm tree can obviously be detected visually in the image, the contrast of the image is significantly lower. This leads to difficulty of distinguishing the young oil palm tree and the background of the image. Thus, the images will be processed using power law transformation to increase the contrast of the images. Several existing and new modified Haar-based rectangular features are introduced to the WorldView-2 image. The features for young oil palm and non-oil palm will be trained and classified using AdaBoost. The resulting classification of young oil palm tree shows a promising result that it can be used for intention of developing automatic young oil palm tree counting.