Forest crown size is a key element related to ecosystem functions, such as timber production and carbon storage. For that reason, there is an increasing interest in forest inventory methods on earth observation data. Base for such mapping are often LiDAR data. Furthermore, spectral information from image data can be used to optimize the mapping.

This project from Christian Weise demonstrate the power of eCognition to extract tree crowns, tree positions, and tree heights using LiDAR data in combination with image data (RGBNir). User will learn how to rasterize point clouds, fill gaps in elevation layers, create individual tree objects using elevation and spectral information, and to convert the objects into 3D vector information to visualize the tree positions, tree trunks, and tree crown diameters.

Important algorithms used in this project:

• rasterize point cloud
• pixel filter 2D (openCV, prototype)
• fill pixel values
• NDSM layer calculation
• watershed segmentation
• convert image objects to vector objects
• vector buffering/shrinking
• create thematic vector object

Download 'Example Project - Tree Crown Delineation'

Interesting related content:

• eCognition Deconstructed: Fill pixel values
• eCognition Deconstructed: Rasterize Point Cloud
• eCognition Deconstructed: NDSM Layer Calculation
• eCognition Deconstructed: Watershed Segmentation
• eCognition Deconstructed: Thematic Layer Operation Algorithms Create / Convert / Remove
• eCognition Deconstructed: Thematic Layer Operation Algorithms Alterations
• Brainwave: Calculating the nDSM
• Project & Rule Set Examples: DTM Modelling of Open-pit Mines