Mapping wild leek with a small drone

We also just got another paper published, in which we outline a methodology for mapping wild leek using a small drone, based on work in Gatineau Park. Wild leek is a small plant that grows on the forest floor, so for most of the year it is impossible to map - in summer it is obscured by the tree canopy, in fall it has already lost its leaves, and in winter it is covered by snow... but for a short period in spring it is bright green while most of the trees it is found under have still not developed their leaves. We flew a drone during that time, processed the imagery through SfM to generate a 3D point cloud, which is pretty standard use of drone imagery. But then we did something really neat - we used an automated algorithm to classify all the points in the point cloud as located either on forest floor or above it, and then we proceeded to eliminate all the points not located on the forest floor. Below you can see an example of the point clouds from two areas, before (left) and after (right) the elimination of the points not located on the forest floor.

This allows us to produce orthomosaics that show what each area would look like when viewed from above, if someone had removed all the trees. In those orthomosaics, green areas by and large correspond to wild leek. We're continuing the research in 2018 by looking at how well this can be done with a camera attached to an airplane rather than a drone (covering larger areas), and also how that compares to using a Sentinel 2 image, which can cover all of Gatineau Park in a single image. so stay tuned...