Remote sensing for reconstruction of natural landscape scenes

In this paper we are going to build the three-dimensional model of the natural scene based on laser scanning data. The complex requirements of plant modeling applications are nowadays supported by high end GIS and remote sensing data gathering methods such as RADAR, LiDAR DTM and DSM data, high resolution satellite data, stereo camera images etc., as well as automated draping technologies for 3D objects. These technologies allow the creation of synthetic and photorealistic 3D landscape visualizations and simulations to display planning alternatives, scenarios and depict clearly their impact on landscape scenery. It is a task that affects both computer graphics and computer vision. The reconstruction of landscape scenes based on the original point clouds from laser scanners is divided into the following subtasks: association of input data into a single coordinate system, LIDAR data classification for reconstruction natural landscape scenes, 3d landscape modeling (the surface of the Еarth), and plant modeling (trees and shrubs). We pose the problem of classification of the original point cloud with LIDAR into three categories: trees and shrubs (woodland), the surface of the ground and other objects (anthropological objects). Highlight one spatial attribute E (x), according to which we will refer to the terms of this or that category. In this work, we present an adaptation of L-systems and Space Colonization algorithm to the Python language, a popular and powerful open-license dynamic language. Visualization of the script is executed by standard tools of the graphical editor Blender. We show that the use of dynamic language properties makes it possible to build complex models.