Model based 3D point cloud segmentation for automated selective broccoli harvesting*

Abstract

The 3D point cloud data was captured in outdoor fields under different weather conditions in 4 locations: 2 in the UK, 1 in Spain, and 1 more in USA using the Kinect 2 sensor. Histograms of the reference models used in our algorithm. A FPFH descriptor [1], based on a set of angular features, is computed for each data point. The descriptor is then matched to both reference models and the difference provides the final classification score. Reference models