Skip main navigation

A guide to 3D data types

A brief guide to the 3D object representation data types: volumes, point clouds and surface meshes
A photo of a flowering plant on a pot alongside a 3D surface mesh of the same plant.

As discussed in the preceding video there are three main types of 3D representations of plants (or other objects), each of which has its own strengths and weaknesses. This article provides a brief reminder and reference guide to volumes, point clouds and surface meshes.

Volumes

A volumetric representation of a plant showing the individual voxels

  • Volumes are 3D arrays of voxels – imagine a stack of cubes like building blocks
  • Each voxel records whether that region of space is occupied by plant material, and possibly how much of that voxel is occupied by material
  • This is a similar principle to 2D image segmentation, though the methods to produce 3D volumes are often different
  • Since there is data for every voxel a key model parameter is resolution
  • Smaller voxels give a higher resolution but there is a computational cost
  • Larger voxels are computationally cheaper but give a blocky appearance and reduces accuracy
  • Volume representations are good for measurement of whole plant traits such as volume, height etc.

Point clouds

A picture of a point cloud 3D representation of a plant, appears as a set of dots in 2D.

  • As the name suggests point clouds are an unstructured set of (x,y,z) coordinates where plant or other material is present
  • As well as coordinates point cloud data may have additional data such as colour attached to each point – especially if the point cloud was created using cameras
  • In contrast to volumes point clouds do not have data at every point in the volume
  • More points means more information but can be difficult to handle, and may increase noise in the data
  • Quality over quantity – with point clouds its better to have fewer good points which you are certain of, than lots of noisy points you are less certain contain of
  • Some traits, e.g. convex hull, can be calculated directly from point cloud data, but most require further processing

Surface meshes

A 3D surface mesh of a plant. The surface is not blocky, like the voxel representation, and appears solid, unlike the point cloud representation.

  • Surface meshes are a set of interlinked triangles that cover (ideally) the entirety of the surface of the plant (or other object)
  • These can be extracted from both volumetric and point cloud data
  • In the case of plants particular care should be taken that the mesh doesn’t link distinct parts of the object.
  • For example, though leaves on different branches may be very close to one another, they should be distinct in the mesh
  • Meshes can be used to make measurements such as surface area, and many others
  • A surface mesh is usually the goal of 3D object representation.
  • Though point clouds can be converted to volumes and vice-versa these are usually stepping stones to produce a final surface mesh, with segmentation and noise removal steps along the way
This article is from the free online

Introduction to Image Analysis for Plant Phenotyping

Created by
FutureLearn - Learning For Life

Our purpose is to transform access to education.

We offer a diverse selection of courses from leading universities and cultural institutions from around the world. These are delivered one step at a time, and are accessible on mobile, tablet and desktop, so you can fit learning around your life.

We believe learning should be an enjoyable, social experience, so our courses offer the opportunity to discuss what you’re learning with others as you go, helping you make fresh discoveries and form new ideas.
You can unlock new opportunities with unlimited access to hundreds of online short courses for a year by subscribing to our Unlimited package. Build your knowledge with top universities and organisations.

Learn more about how FutureLearn is transforming access to education