Tree Detection

Canopy Height Model

In addition to DTM and DSM, literature also defines Canopy Height Model (CHM). The CHM is the difference between the DTM and the DSM over a forest zone.

Tree detection

Tree detection is the process of identifying the location of trees in a forest. To find the position of individual tree, we can use different approaches.

Tree location algorithms assume that the trunk of the tree is located under the highest point of its crown. The crown is the part of the tree that is above the ground. Then finding tree locations is equivalent to finding local maxima in the CHM.

The local maximum depends on the size of the window used to find it.

This parameter which have a huge impact on the quantity of tree detected can be, depending on the method, set by the user or automatically computed.

An allometric equation linking the tree height to the diameter of the crown can be used to compute the size of the window. The parameter of the equation can be estimate by detected some tree and measuring their height and crown diameter, the equation is then derive from a linear equation.

Tree segmentation

Raster-base approach

Dalponte

The method proposed by Dalponte et al. (2016) [DD16] is based on the following steps:

  • Low-pass filtering of the CHM

  • Local maxima with circular moving window

  • Region growing from local maxima

Silva

The method proposed by Silva et al. (2016) [SMD16] is based on the following steps:

  • Detection of local maxima

  • Voronoi tessellation

  • CHM clipping