You are here

Conifer Species Classification Approach


Project researchers

The Airborne light detection and ranging (LiDAR) is a popular active remote sensing technique that over-samples the remote study area using a highly directed laser beam, resulting in the generation of a very dense georeferenced elevation points called as the LiDAR point cloud. Unlike other visible remote sensing techniques, the data aquired using high sampling-density small-footprint muti-return LiDAR can record  a huge amount of  structural information about the forest vertical profile. This inforamtion can be used to study forest structures, low vegitations and even the ground beneath the canopy.

The knowledge about the species of trees is essential for precision forest management practices. Modern high density airborne Light Detection and Ranging (LiDAR) systems have the ability to acquire large number of LiDAR points, allowing a very detailed characterization of the forest at the individual tree level. In this context, it is possible to use LiDAR data for accurate classification of the tree species. We consider the specific problem of species classification of trees belonging to the conifer class. Conifer species classification is particularly challenging when only the external crown geometric information is considered. To address the problem we propose a novel approach that model the internal crown structure of the conifers. The internal structure is identified by using 3D region growing and Principal Component Analysis (PCA) and is used for defining a set of novel Internal Crown Geometric Features (IGFs).

Research topics: