05. Basic Workflow
Once the application is installed, the workflow is to set up a project, import data, and segment data to terrain, vegetation, and trees. Only then the tree and crown features can be evaluated.
The first step when using the 3D Forest is to set up a project. This can be done via the Menu -> Project -> New Project. In this step, a directory of the whole project can be selected. This directory then becomes a place for data storage. The transformation matrix, i.e. the data extension, can be set at this point.
Once the project is set, the raw data can be imported. It is essential to import them as the base cloud if they consist of all points of the forest (terrain, trees, buildings, etc.). In case the data was already separated into terrain and vegetation, each point cloud has to be imported separately according to its type (terrain or vegetation). Likewise, the clouds of individual trees should be imported as tree clouds.
It is important to check the cloud quality after the import. Mainly if there are any doubts during setting-up a project. If the cloud looks like parallel lines, the transformation matrix is wrong. Then the project has to be recreated with a different (appropriate) transformation matrix.
Once the base cloud is imported, it needs to be divided into two basic parts: terrain and vegetation. Two alternative methods are provided to fit this request. Terrain from octree, which returns ground points of cloud. And terrain from voxel, which gives a voxelized cloud represented by centroids of original points in each voxel. All points that do not belong to the terrain cloud can be manually selected and removed during a visual check. You can also use some of the implemented filtering methods.
There are functions to clean terrain point cloud using the following filters: statistical outliers filter and radius outlier filter provided by PCL. Besides, there are functions for interpolation of missing terrain using inverse distance weighting (IDW) if gaps are present, and functions for evaluation of slope, aspect, curvature, hillshade, and a function that helps to identify features within the terrain point cloud.
For a tree analysis, it's essential to work with clouds that represent single trees. This is achievable by automatic or manual segmentation of the vegetation. During manual segmentation, all the points that do not belong to the tree have to be removed before saving a single tree cloud. On the contrary, automatic segmentation treats the whole vegetation cloud according to the input parameters. A more detailed description of the automatic segmentation method is presented in section Vegetation.
If needed, the tree cloud edit helps to manually remove points that do not belong to the particular tree. It also enables to merge a tree separated into two clouds.
The 3D Forest is designed for individual tree analysis. Thus the first attribute to be evaluated is the tree position (defined as the tree base position). This is required for the estimation of all other tree variables like tree DBH, height, length, stem curve, or tree planar projection.
TLS is a great tool for scanning objects in remote distances, e. g. a tree crown. However, the analysis of tree crown requires the position and tree height to be calculated first. Unless this is done, the crown analysis menu is inactive in the Crowns Menu (greyed out). Besides, the crown analysis has to be computed using the cloud that consists of points that belong to the tree crown only, i.e. not to the stem. There are both manual and automatic segmentation procedures to do that. Once the crown is segmented from the tree, crown features (crown height, crown length, crown width, crown centroid, crown displacement, crown voxel volume, crown convex 3D hull, crown concave hull, or intersection of crowns) can be evaluated.
The last group of attributes is listed under the tree quantitative structure model (QSM). It gives information about stem and branches (length, connection). As well as for the crown analysis, the position and tree height has to be evaluated first.