Point clouds have exploded onto the construction industry, offering a revolutionary solution for capturing and analyzing three-dimensional data. With their ability to create accurate and detailed models of sites and structures, point clouds are changing the way construction projects are planned, designed and executed. 

This technological breakthrough is driving efficiency, accuracy and collaboration in the industry, opening up a new range of possibilities for improving quality, reducing costs and minimizing risk on construction sites.

What is a point cloud?

A point cloud is a three-dimensional representation of an object, environment or surface, composed of a large number of points in space. These points are obtained by technologies such as laser scanning, terrestrial scanning or photogrammetry, which capture the position and coordinates of each point relative to a reference system. 

Each point in the cloud contains spatial information and, in some cases, additional attributes such as colour or intensity. By joining all the points together, an accurate and detailed digital representation of the scanned object or environment is created, allowing it to be visualized, analyzed and manipulated in virtual environments or specialized software. 

Point clouds are used in a variety of fields, including architecture, engineering, construction, mapping and heritage conservation, among others, due to their ability to capture three-dimensional data with high accuracy.

How to classify point clouds

Classifying a point cloud is a fundamental process to extract information and better understand the captured data. These are some common techniques used to classify point clouds:

  1. Intensity-based: Some laser scanning and photogrammetry devices capture the intensity of the light return at each point. This information can be used to classify points according to their reflectivity or surface characteristics, such as distinguishing between vegetation, asphalt or buildings.
  1. Geometry-based: Geometry-based classification uses the position and spatial relationships of points to identify specific objects and features in the cloud. This is achieved by algorithms that analyze the density, shape and structure of points to identify surfaces, edges or abrupt changes in terrain.
  1. Colour-based: If the point cloud contains colour information, it can be used to classify points based on their chromatic appearance. This can be useful in applications such as urban mapping, where different building materials can be distinguished or objects can be identified based on their characteristic colour.
  1. Based on additional features: Depending on the technology used to capture the point cloud, additional information such as laser return intensity, temperature or texture may be available. These features can be exploited to classify points according to specific properties relevant to the application in question.

Point cloud classification can be performed manually, where an operator manually assigns labels to points, or by algorithms and machine learning techniques that automate the process. In both cases, classification is essential to effectively analyze and use the captured data in applications such as design, inspection or terrain analysis.

 Point clouds: How data is obtained

Point cloud data are obtained by means of three-dimensional capture technologies, which allow the collection of accurate information about the geometry and characteristics of an object or environment. Some of the most commonly used technologies to obtain the data are:

  1. Terrestrial laser scanning (TLS): This technique uses a laser scanner mounted on a tripod or vehicle to emit laser pulses into the environment. The scanner measures the time it takes to receive the reflected pulses, which allows the distance between the scanner and surrounding objects to be calculated. By scanning a large area and combining multiple scans, a detailed three-dimensional point cloud can be generated.
  1. Terrestrial photogrammetry: Photogrammetry uses photographs taken from different angles to calculate the three-dimensional geometry of an object or environment. By using algorithms and feature matching techniques, 3D points can be extracted from the captured images. By combining measurements from multiple images, a three-dimensional point cloud is generated.
  1. Airborne laser scanning (ALS): This technique involves the use of laser scanners mounted on aircraft or drones to capture data from the air. Laser pulses are emitted towards terrain and objects, and the scanner records the time it takes to receive the reflected pulses. This allows the generation of a point cloud with high coverage and resolution.
  1. Structured light scanning: This technique uses a projector that emits a structured light pattern, such as lines or dots, onto the object or environment. A camera captures the deformations of the light pattern due to the geometry of the object, and by analyzing these deformations, 3D points are generated to form the point cloud.
  1. Sensor-based capture: There are also specialized sensors that can be used to capture 3D data. For example, time-of-flight (ToF) sensors emit pulses of light and measure the time it takes to return, allowing distances to be calculated and the point cloud to be generated.

These are just some of the most common technologies used to obtain point cloud data. Each has its own advantages and technical considerations, so the choice of technique depends on the specific application and project requirements.

Point cloud to mesh

Point cloud to mesh is the process of converting a point cloud into a three-dimensional mesh. These points are obtained by 3D scanning techniques, such as laser scanning or photogrammetry, and contain information about the shape, position and density of the scanned objects.

The conversion of a point cloud into a three-dimensional mesh involves the creation of a solid and continuous surface structure that resembles the objects represented by the points. A three-dimensional mesh is composed of vertices (points in three-dimensional space), edges (connections between vertices) and faces (surfaces formed by the edges).

This process involves using algorithms and techniques to reconstruct the underlying geometry from the points in the cloud. These algorithms can perform operations such as triangulation or fitting parametric or implicit surfaces to the points to create a smoother and more accurate representation of the object.

Some key steps in performing the conversion are:

  1. Point cleaning and filtering: Prior to conversion, it is essential to perform a cleaning and filtering process of the point cloud to remove outliers, noise or scanning errors. This ensures a more accurate representation of the object.
  1. Sampling and discretisation: Depending on the density and size of the point cloud, sampling or discretisation may be necessary to reduce the number of points without losing too much important information.
  1. Triangulation: One of the most common approaches to convert a point cloud into a mesh is by triangulation. This process involves generating connected triangles between neighbouring points, creating a mesh structure.
  2. Surface fitting: In some cases, it is possible to use surface fitting techniques to obtain a more accurate representation of the object. This involves fitting curves and parametric surfaces to points in the cloud, creating a continuous surface.
  1. Optimisation and refinement: Once the initial mesh has been generated, optimisation and refinement techniques can be applied to improve the quality of the mesh, such as smoothing the surface, removing imperfections or reducing the number of polygons.

Once the point cloud to mesh conversion is done, a more structured and visually understandable three-dimensional representation of the scanned object is obtained. This is useful in a variety of applications such as visualization, 3D modeling, animation, simulation and reverse engineering.

Point clouds are extremely important for this sector due to their ability to capture accurate three-dimensional data of sites and structures. In addition, this data allows for detailed analysis, more efficient planning and early detection of problems, resulting in greater efficiency, quality and accuracy.