Advanced remote sensing

  1. LiDAR techniques and applications

Purpose: using the available .las file perform extraction of the vegetation mask as a raster using eCognition software package.

So the first step was converting point cloud .las file to raster image. In the work, Number of returns as point property were used in order to extract Vegetation Mask – so a significant amount of the LIDAR energy can penetrate the forest canopy like sunlight. A massive number of points LIDAR can receive. Below there is a picture of a tree, shown the number of returns meaning laser pulse hits multiple times the branches of a tree and returns back to the sensor.


The reflections are received from the different parts of the tree – 1st, 2nd, 3rd returns until it finally hits the bare ground – Last Return. If there’s no forest in the way, it will just hit the surface. As with the case of trees, one light pulse could have multiple returns.  The first returns used to get the top of canopy, where last return used to define the ground by converting .las file to raster image. Further on the new raster image the pixels with several (multiple) returns are shown as vegetation and one return is ground or solid areas as roofs of buildings, ground, roads and etc. So last step was just subtract the raster images of first return and last return (ground) in order to define areas with number of return > 1 or to extract Vegetation, which has multiple returns. Below there are a rule set for the point cloud data and resulted image with vegetation: