What is a point cloud? 3 steps for beginners to grasp it quickly | Uses, how to create, and application examples

Contents

• Step 1: What is a point cloud?

• Step 2: How to create point cloud data

• Step 3: Uses and applications of point cloud data

• Summary

In recent years, the term "point cloud" has become increasingly common in the construction and civil engineering industries. This is because the use of point cloud data has expanded with the proliferation of 3D laser scanners and the Ministry of Land, Infrastructure, Transport and Tourism's promotion of i-Construction, along with the digitization of construction sites. Point clouds can digitally record site conditions with extremely high accuracy, making them of interest in a wide range of fields such as assessment of infrastructure deterioration, maintenance management, and disaster prevention measures. However, many people may still wonder, "What exactly is a point cloud?" and "What can it be used for?" In this article, to help beginners understand point clouds as quickly as possible, we explain the basics in three steps. We will cover the basic concepts of point clouds, how the data is created, and examples of applications in order.

Step 1: What is a point cloud?

Point cloud data is, as the name implies, an aggregate of many "points" in space—three-dimensional data that represents the shape of an object (called "Point Cloud" in English, named for the way countless points gather like a cloud). Just as a photograph is composed of tiny dots that form an image, a point cloud records the shape of 3D space through a multitude of points. For example, when a building or terrain is measured with a laser scanner, countless points on the surface are captured and displayed on a computer like a 3D photograph made of points. This enables the site to be preserved as digital data that includes depth information that conventional plans or photographs could not fully capture.

In other words, point cloud data is a highly accurate digital copy of the physical space as a whole. It can survey wide areas at once and capture detailed geometry down to millimeter-level (mm (0.04 in)), and it also makes it easy to measure required dimensions later or extract arbitrary cross-sections. Whereas conventional manual surveying involved measuring point by point and drafting drawings, point cloud measurement can cover large areas in a short time with almost no missed measurements. Because the acquired site conditions can be converted directly into 3D data, it is possible to create accurate 3D models and drawings afterward even for structures that do not have as-built drawings. Thus, point cloud data is expected to become a powerful tool for construction management and maintenance.

Note that alongside point cloud data you will often hear the term "LiDAR." LiDAR is the name of a remote sensing technology (a 3D measurement method) that uses laser light, while point cloud data refers to the three-dimensional measurement data obtained by such measurement techniques. The two are often confused, but keep in mind that LiDAR is simply the name of the technology, and point cloud data are the data produced as a result.

Each point that composes a point cloud contains the basic elements of positional coordinates (X, Y, Z), and depending on the method, attributes such as color information (RGB values) or laser return intensity may also be added. For example, point clouds generated by photogrammetry are assigned colors based on the captured images, and point clouds from laser scanner measurements may record the intensity of the laser reflected from the instrument as well as classification information. Because they consist of an enormous number of points, the data volumes are very large: files on the order of millions of points can range from hundreds of MB to several GB, and point clouds covering an entire city can reach hundreds of GB. Therefore, it should be noted that handling point cloud data requires high-performance computers and dedicated software environments.

Step 2: How to Create Point Cloud Data

Next, let's look at how this point cloud data is acquired and created. There are, in fact, several ways to obtain point clouds, using specialized 3D measurement equipment and image analysis techniques to capture three-dimensional data on site. Broadly speaking, there are two approaches: an approach that uses laser light to directly measure distances (utilizing LiDAR technology), and an approach that reconstructs three-dimensional shapes from photographs (photogrammetry). Various applied methods for each are used in the field. The main methods include the following.

Measurement using ground-based 3D laser scanners: This method involves setting up a tripod-mounted laser scanner on site and emitting laser light 360° around to acquire point clouds. It rapidly calculates distances to objects from the round-trip time of the laser and can record several million coordinate points per second. Because it can convert the shapes of buildings and civil engineering structures into high-density 3D data in a short time, it is suitable for detailed measurements requiring millimeter accuracy. Typically, when measuring interiors of buildings, the scanner is set up in each room and measurements are taken several times; the multiple point clouds acquired are later aligned and integrated into a single 3D model. In recent years, laser scanners equipped with a 360° camera that can add color information to point clouds using photographic images have become common. Currently, ground-based laser scanners are the most common point-cloud acquisition method in civil surveying and construction quality control.

Point cloud generation by photogrammetry (photogrammetry): a technique that reconstructs the three-dimensional shape of a target from a large number of photographs taken with a camera. By photographing buildings or terrain from various angles and computing correspondences of feature points between images, point cloud data is generated. This includes large-scale 3D reconstruction of terrain using aerial photographs taken by a drone (UAV) and measurements by SfM (Structure from Motion) analysis using a DSLR camera on the ground. Photogrammetry can be easily applied to wide-area terrain surveying and construction record keeping, and recent advances in software have made it possible to obtain point clouds with higher accuracy and resolution. To provide point clouds generated from photographs with accurate dimensional scale and coordinates, calibration work such as placing known points on the ground and combining with high-precision GPS positioning is important. For example, by shooting with a drone equipped with RTK-GNSS, it has become possible to assign centimeter-level (half-inch-level) position information to each photo and generate point clouds in a map coordinate system.

Drone-mounted LiDAR surveying: This method equips a drone with a compact LiDAR sensor (laser scanner) to perform direct laser measurement from the air. Because it can scan the ground surface from the air, it is effective in locations that are difficult to measure from the ground, such as terrain hidden beneath forest trees or steep mountainous areas. In addition, because wide areas can be surveyed at once in a short time, the acquired point cloud data can be used to calculate earthwork volumes (cut-and-fill) and to assess disaster conditions. There are also cases where airborne laser survey data provided by the Geospatial Information Authority of Japan and local governments are offered as open data, and by reusing existing point cloud data it is possible to reduce the effort required for on-site surveying. Furthermore, as LiDAR sensors become smaller and lower-cost, drone-based laser surveying is expected to become increasingly common.

Mobile Mapping Systems (MMS): There are systems that mount multiple laser scanners, cameras, and GNSS units on automobiles or rail vehicles to continuously acquire surrounding 3D data while moving. Because they can efficiently survey long-distance infrastructure such as roads and tunnels, their use has advanced for detailed understanding of road geometry, tunnel displacement measurement, and road asset management. For example, a dedicated survey vehicle can drive along trunk roads while scanning the pavement and surrounding structures to generate point clouds, which helps update road registers and detect pavement damage. Because MMS can obtain high-density point clouds even while moving at high speeds, it has attracted attention as a surveying technology in recent years.

Measurements with handheld scanners and smartphones:

Small, portable 3D scanners that can be carried by hand and point-cloud measurement methods using commercially available smartphones and tablets have also emerged. Handheld LiDAR scanners use SLAM (simultaneous localization and mapping) technology, allowing an operator to acquire surrounding point clouds in real time simply by walking around. In addition, recent smartphones (for example, some higher-end iPhone models) are equipped with LiDAR sensors, and with dedicated apps you can easily 3D-scan familiar spaces. Point-cloud measurement with a smartphone is attractive for its mobility and ease of use, and has begun to be used for measuring tight indoor spaces and small structures, and for recording construction progress. However, because the measurement accuracy and effective range of built-in smartphone LiDAR are limited compared with dedicated equipment, they are currently mainly used for simple measurements of small-scale targets.

There are many diverse ways to acquire point cloud data, each differing in accuracy, applicable range, and required equipment. Terrestrial laser scanners offer high accuracy and excel at detailed measurements, while drone photogrammetry is suitable for quickly recording wide-area terrain. Drone-mounted LiDAR and MMS are effective for comprehensively capturing infrastructure, and smartphone and handheld systems have strengths in mobility and ease of use. By selecting and combining these methods according to on-site needs, efficient and highly accurate point cloud data acquisition can be achieved.

Step 3: Applications and Use Cases of Point Cloud Data

Finally, let's look at how the acquired point cloud data can be used on site. Point cloud data is increasingly being put to practical use in various aspects of civil engineering and construction. It brings value that could not be obtained from traditional 2D drawings or single-point surveying, and is expected to serve as a tool to drive on-site DX (digital transformation). Point cloud data is truly a "game changer" that digitizes entire sites and enables multifaceted utilization. Here are examples of the main application areas.

Current condition assessment and preparation of design documents: By using point clouds, you can accurately record the existing conditions of structures and terrain and create detailed 3D models and 2D drawings from them. For example, even when original drawings are not available for renovation designs of old bridges or tunnels, it is possible to generate accurate plan views and cross-sections from point clouds obtained by scanning the site. The acquired point clouds can be used in dedicated software to extract the required cross-sections or imported into CAD or BIM software as the basis for drafting. Condition assessment using point clouds greatly helps improve design accuracy and streamline planning and review processes.

As-built management (post-construction inspection and verification): After construction is completed, structures and developed ground can be measured with point clouds, and by comparing the as-built condition to the design model and drawings, the finishing accuracy can be verified in detail. By measuring member dimensions and slopes from point clouds and visualizing the differences from the design data as a color map, even minute deflections of concrete surfaces can be detected. In a case study of a tunnel project, using point clouds to check the as-built condition of rebar reportedly reduced work time and costs by about 70% compared with traditional manual methods. As-built management using point clouds leads to early detection of construction errors and reduced rework, contributing to both quality assurance and improved efficiency.

Visualization of construction progress and work-quantity management: If a construction site is scanned regularly and converted into point clouds, the progress of the work can be visualized in 3D. For example, if drone surveys are conducted weekly at an earthwork site and the resulting point cloud data are overlaid and compared week by week, you can immediately see how far excavation and embankment have progressed. Furthermore, by calculating cut-and-fill volumes from the point cloud, daily quantities of work-in-progress can be measured accurately, and the preparation of progress-quantity management documents can be streamlined. Quantitative assessment of construction progress, which was difficult with only photographs or visual inspection, becomes easy, helping to advance schedule management. Visualizing progress with 3D data also smooths information sharing with owners and stakeholders and enables objective progress evaluation.

Earthwork volume calculation and disaster response: By calculating soil volumes from acquired high-resolution terrain point clouds, you can derive volumes with far greater accuracy than those traditionally estimated from only a few dozen survey points. The use of point cloud data is advancing in disaster prevention as well—not only improving the accuracy of site quantity control, but also enabling rapid estimation of the volumes of soil washed away or collapsed during disasters.

Infrastructure inspection and maintenance: Point clouds are also being used for the upkeep of infrastructure structures such as bridges and tunnels. These include regularly scanning tunnel interiors to monitor long-term changes in 3D, and color-coding differences between bridge point-cloud models and historical data to detect areas of deterioration. Also, if point clouds of structures captured at completion are archived as digital data, they can be reused repeatedly for future repair design and for comparing damage after disasters. Point-cloud data become detailed long-term records of the site and serve as foundational information for the so-called "digital twin," aiding maintenance and future planning. By utilizing such 3D data, infrastructure inspection tasks that traditionally required significant manpower and time are streamlined, and even minute changes can be detected and evaluated quantitatively.

Other applications of point cloud data include construction planning simulations, safety management, cultural heritage preservation, and the entertainment sector. For example, in construction planning, the working range of heavy machinery can be simulated on point clouds to help plan crane installation on confined sites. In safety management, dangerous areas that people cannot enter can be scanned by drones to remotely ascertain conditions, which helps reduce workers' risk. In the cultural heritage field, ruins and historic buildings can be high-precision 3D scanned and digitally preserved so that detailed records remain even if the originals are damaged or lost. The acquired data can also be used for virtual exhibitions or as reference material for restoration work. In the entertainment industry, real streetscapes can be converted into point clouds for use as CG backgrounds in games and films, or used as foundational data when overlaying digital information onto real space in AR apps. The range of applications for point cloud data is expanding year by year, and integration with AI (artificial intelligence) technology is expected to enable advanced analyses such as automatic recognition and modeling of terrain and structures from point clouds. As measurement technologies become more accessible in the future, point clouds are expected to be utilized in all aspects of on-site work, becoming the foundation that supports the digitization and advancement of operations.

To make the most of acquired point cloud data in practical work, appropriate post-processing and editing are also indispensable. You perform alignment (registration) to integrate point clouds measured from multiple locations into a single coordinate system, remove noise points, and thin the data (decimation) as needed to make it manageable. If the data volume is very large, it is also necessary to split files or apply effective compression. Furthermore, specialized tools are used to generate surface models (meshes) for terrain representation or line data from point clouds, and to create drawings in CAD software. In recent years, advances in point cloud processing software have enabled advanced functions such as automatically removing unwanted objects or extracting only the ground surface to generate terrain models in TIN format. By obtaining high-precision 3D data through such post-processing, it becomes much easier to use it for design and construction simulations and various analyses. The high-quality point cloud data obtained in this way can serve for a long time as foundational information for building a site's digital twin and for future maintenance management.

Summary

Point cloud data is bringing significant changes to field recording and management methods. It may seem difficult at first, but if you understand the basics and start using it little by little, you will come to appreciate its convenience and value. In fact, you do not need to purchase an expensive set of 3D surveying equipment from the outset; it is effective to begin with small-scale measurements using a smartphone or an inexpensive handheld scanner. Even creating point clouds of small objects lets you experience their usefulness, making it easier to step up gradually to full-scale deployment. In addition, a variety of viewers and analysis software for handling point cloud data are available. For example, there are free point cloud viewers and open-source processing software, allowing you to try viewing and editing point cloud data at low cost. As needed, you can make use of surveying equipment rentals or adopt a phased introduction by initially outsourcing only the measurements to a specialist and learning how to utilize the data.

Fortunately, thanks to advances in technology, easy-to-use point cloud measurement tools that even beginners can handle have emerged. For example, by using the LRTK Phone, a GNSS high-precision positioning device that attaches to an iPhone, you can easily start point-cloud surveying with centimeter-class accuracy (cm level accuracy (half-inch accuracy)) using only a smartphone. In addition, LRTK Phone automatically attaches positional information such as latitude, longitude, and elevation to the acquired point cloud, and you can instantly measure distances, areas, and volumes within the dedicated app. Because it allows intuitive 3D measurement even without specialized knowledge, it is attracting attention as a new "weapon" on site. By effectively incorporating these latest tools, try to make 3D point cloud technology your ally in a way that fits your company’s operations. We hope that the utilization of point cloud data will lead to improved productivity and the promotion of DX at your sites. Why not make good use of point cloud data and take your on-site operations to the next stage? That concludes the three-step explanation from the basics of point clouds to examples of utilization. We hope it will be helpful for your future work. Thank you for reading.