Arbor canopy extraction method and system based on unmanned aerial vehicle and radar and medium

The invention provides an arbor canopy extraction method and system based on an unmanned aerial vehicle and a radar, and a medium. The method comprises the following steps: filtering original three-dimensional point cloud data collected by the unmanned aerial vehicle; performing rasterization processing on the filtered three-dimensional point cloud data; obtaining gradient raster data through a gradient calculator; segmenting the gradient raster data through an image segmentation model, and obtaining arbor canopy data from a segmentation result; carrying out post-treatment filtering; according to the design, two-dimensional canopy height data is replaced with gradient raster data information, dependence on ground points is eliminated, gradient information is calculated directly through a digital surface, operation and calculation of any ground points are not involved, the problems of ground point missing and large ground point recognition calculation amount are effectively avoided, and the calculation efficie