A point cloud survey is a process that uses technologies like laser scanning or photogrammetry to capture vast amounts of 3D data from a physical object or space. The output is a "point cloud," a collection of millions of data points, each with a precise X, Y and Z coordinate. Its main benefits are its speed and the massive amount of detail it captures. However, for the technology to be truly useful, it must also be accurate, a key factor in deciding whether the technology is suitable for a specific job.
A point cloud survey is like taking a very detailed picture of a room with millions of tiny dots, each perfectly placed. The reason professionals use this technology is because it's much faster than measuring everything via traditional survey techniques or even by hand and provides an incredible amount of information. But if those points are not in the right places, or if they are a little bit off, the picture is not useful for building something precisely. So, while speed and detail are great, the real value of the technology comes down to how accurate it is.
The potential for sub-centimetre accuracy is often cited as the reason for using the technology, but real-world results vary. The most significant factors that affect a survey's precision include the equipment used, the environmental conditions and the complexity of the object being scanned. So, let’s first take a look at the equipment.
Equipment and methodology
• Terrestrial Laser Scanners (TLS): These are typically the most accurate, capable of capturing data with a precision of a few millimetres. They are fixed on a tripod, which eliminates movement and allows for dense, high-quality scans. However, even with TLS, the final federated point cloud (the result of stitching together multiple scans) can have errors if the alignment, or "registration", is not with set acceptable tolerances.
• Photogrammetry: This method uses a series of overlapping photographs to create a 3D model. Its accuracy is generally estimated to be a few centimetres, making it suitable for larger-scale mapping or less intricate projects where a high level of detail is not the primary requirement.
• Mobile and Drone-Based Scanning: While offering incredible speed and coverage for large areas, handheld mobile scanners and drones generally have a lower accuracy tolerance compared to a static TLS.
Environmental and on-site factors
As well as the equipment being used, the material, colour and reflectivity of the surfaces being scanned can also have a significant impact on accuracy. Dark or highly reflective surfaces, like glass or polished metal, can be challenging for lasers to scan, leading to "noise" or missing data.
Atmospheric conditions such as rain, fog, or even strong sunlight can also interfere with the laser beams, reducing the quality and density of the data collected. Furthermore, the distance and angle of the scanner to the object are crucial. The greater the distance from the scanner to the object can result in less dense data and a higher potential for error. The scanner’s position and density settings are choices made by the user that can affect the quality and accuracy of the output.
Accuracy in practice
While a terrestrial laser scanner can achieve sub-centimetre accuracy in a controlled laboratory setting or for small, specific sections of a project, a sizable point cloud of a large, multi-story building may show registration errors in the range of a few centimetres. These systematic errors, which can be hidden from visual inspection, can cause misalignments between different portions of the point cloud. Therefore, it is critical to perform an independent validation, such as a total station control survey, to verify the data's accuracy before it is used for downstream applications like BIM (Building Information Modelling).
A point cloud's accuracy is best measured against the specific project's requirements. For intricate architectural details on a heritage building, for example, a millimetre-level of detail is essential. However, for large-scale topographical surveys or preliminary site planning, a centimetre-level of accuracy may be sufficient. The level of detail and accuracy required directly influences the time spent on-site, the equipment used and the subsequent data processing, all of which impact the project's overall cost and timeline.
A point cloud survey offers a significant leap forward in precision and efficiency over traditional manual methods. It virtually eliminates the need for multiple site visits and reduces the risk of human error in data collection. However, its accuracy is not absolute. A successful survey relies on a skilled operator who understands the interplay of equipment, environment and geometry to produce a data set that is not only highly detailed, but is also fit for its intended purpose.
