Predicting Important Commercial Timber Species Using Drone Aerial Image

Abstract

Identification of tree species prior to logging is one of new initiatives to support the management of tropical forest. In this project, the use of drone and the improved technique of image processing will increase the potential of identifying tree species. only when the value of timber is found to be attractive for timber operation, then the area could be opened for timber operation. otherwise that forest area could be assigned for protection of biodiversity and other environmental purposes. this initiatives will save the cost of unnecessary ground inventory and uneconomic timber operation. in the analysis of the images, several algorithm are used for the segmentation and identification of species through crown images. the approach is effective as the technique will focus on the commercial timber species which could be easily seen on the top layer of the forest landscape. based on our preliminary result, the technique could achieve about 60% - 70% accuracy of the segmentation and the prediction of commercial timber species. the success of this initiatives could be provide the first phase of the forest inventory which will give tremendous impact to the feasibility of logging operation and sustainable forest management for the forest area.