Capture on-site and simply send it to the cloud | How to automate point cloud generation to streamline surveying operations

Table of Contents

• What is point cloud data? The traditional point cloud creation process and its challenges

• A new point cloud generation workflow: capture and upload to the cloud

• How automatic point cloud generation becomes possible in the cloud

• Benefits of using the cloud to improve surveying work efficiency

• Use cases of cloud-based automatic point cloud generation

• Key points to consider when implementing

• Simplified surveying with LRTK

• FAQ

In field surveying, land and structure point cloud data (three-dimensional data composed of a large collection of points) is essential information for design and construction management. However, traditionally, acquiring and processing point clouds required specialized equipment and advanced PC software, and data processing involved significant effort and time. With the emergence of services that can automatically generate point cloud models just by capturing on site and uploading to the cloud, the surveying field has begun to change dramatically. In this article, we explain in detail the methods and effects of automating point cloud generation via the cloud. We also introduce the efficiency benefits and specific use cases gained from implementation, and finally present simple surveying with LRTK that enables anyone to easily perform 3D surveying.

What is point cloud data? Conventional point cloud creation process and challenges

First, let’s briefly review what point cloud data is. Point cloud data is three-dimensional survey data consisting of countless points acquired by technologies such as laser scanners and photogrammetry (photo surveying). Each point contains X, Y, Z coordinates (positional information), enabling high-precision representation of the shapes of objects and terrain. For example, in civil engineering and construction, point clouds—which can digitize terrain undulations and the shapes of structures in their entirety—are widely used for tasks such as checking against design drawings, volume measurement, and construction progress management.

Traditionally, there have been two primary methods for acquiring point cloud data on site. One is to directly measure point clouds using ground‑based or mobile laser scanners, and the other is to capture numerous photos with drones or DSLR cameras and reconstruct three-dimensional data (photogrammetry) using specialized software. However, these conventional point cloud generation processes had several challenges.

• Expensive equipment and software required: High-precision laser scanner equipment is extremely costly, and the initial investment was a major burden. Photogrammetry software is also often specialized and expensive, and without a high-performance PC processing could not be carried out smoothly.

• Data processing is time-consuming and labor-intensive: When generating point clouds from photos, aligning hundreds of images and performing 3D analysis took a long time. Dedicated technicians had to operate the software and adjust parameters, and in some cases processing taking more than a night was not uncommon.

• There is a time lag before results are available: It could take days from acquiring data on site to completing the point cloud model, during which the conditions at the site might change. Even for urgent surveys it was difficult to obtain results the same day, causing on-site responses to fall behind.

• Specialized knowledge required: Creating and processing point clouds required specialized skills and was not something anyone could handle casually. As a result, the workload tended to concentrate on a limited number of technicians.

Thus, while acquiring and generating point cloud data is extremely useful, it has posed significant hurdles in terms of effort and cost.

A new point cloud generation workflow where you just shoot and upload to the cloud

In recent years, a new workflow has emerged that has completely changed this situation. It is the use of cloud services where point cloud generation is completed simply by uploading the data captured on site to the cloud. On site, you simply photograph the target area as usual with a smartphone, digital camera, drone, etc. After shooting, if you send (upload) the resulting photo data to the cloud, advanced processing is automatically performed on the backend and a three-dimensional point cloud model is generated.

With this cloud-based automated processing, steps that used to be done manually—image alignment, feature-point extraction, and 3D reconstruction—are all handled by the service, freeing users from cumbersome operations. It truly makes "point cloud generation simply by uploading to the cloud" possible. On-site personnel only need to send the data; they don't have to carry heavy laptops or rush to perform processing on-site.

For example, consider wanting to measure the volume of backfill for foundation work at a construction site. Traditionally, a surveyor would perform detailed measurements on site, bring that data back to the office, and then convert it into a point cloud and calculate the volume. By using a cloud-based point cloud generation service, you can simply fly a drone over the site, upload the aerial photos to the cloud on the spot, and obtain a 3D model and automatic soil-volume calculation results in a short time. Processing can proceed in the cloud while you travel back from the site to the office, so by the time you arrive the results may already be ready. In this way, the simplicity of completing the task with only shooting and uploading dramatically increases the speed and efficiency of surveying.

How automatic point cloud generation in the cloud works

So, how are point clouds generated just by uploading photo data to the cloud? Behind this are advances in photogrammetry and cloud computing. Servers in the cloud run advanced image analysis algorithms that extract feature points from multiple submitted images, match them against each other, and recover the camera positions and orientations along with three-dimensional coordinates. Using methods commonly called "Structure from Motion (SfM)" and "Multi-View Stereo (MVS)", the 3D shape of the subject captured in the photos is reconstructed and output as point cloud data.

This process consumes a large amount of computer resources, but because the cloud service performs GPU-accelerated parallel processing, results can be obtained faster than before. Also, if the photographic data includes location information such as GNSS mounted on a drone or survey coordinates acquired on the ground, that information is automatically read and reflected in the point cloud, producing a point cloud with coordinates that aligns with the real-world coordinate system. Some services also offer advanced features in which AI, when necessary, removes unwanted noise from images and point clouds or automatically differentiates and classifies ground surfaces and structures.

Furthermore, point cloud data captured by LiDAR-equipped smartphones and simple 3D scanners—which have become increasingly common in recent years—can be uploaded to the cloud and integrated with other photographic data to generate models. For example, by merging in the cloud a point cloud scanned for fine details with a smartphone’s LiDAR and a wide-area point cloud generated from drone aerial photographs, you can obtain a 3D model that reproduces both extensive areas and intricate details with high fidelity. In this way, the cloud combines various data sources and continuously performs point cloud generation using the latest and most appropriate algorithms.

Benefits of Streamlining Surveying Operations by Utilizing the Cloud

Introducing cloud-based automatic point cloud generation brings numerous benefits to surveying operations. Here, we will look at the main advantages.

• Significant time savings and immediacy: Automated processing dramatically shortens the time from on-site work to deliverable completion. Point cloud generation, which used to take several days, can in some cases be completed in a matter of hours thanks to high-performance cloud processing. As a result, survey results can be shared and reviewed on the same day, accelerating construction decision-making.

• No need for specialized equipment or high-performance PCs: With cloud services, all that is needed on-site are consumer digital cameras, smartphones, drones, and an internet connection. There is no need to purchase expensive laser scanners or workstation-class PCs for point cloud processing, significantly reducing upfront and maintenance costs. Results can be checked from mobile devices, offering the convenience of use regardless of location.

• Reduced workload and human error: Automating complex processing steps reduces the workload on personnel. Concerns about human errors such as software operation mistakes or calculation errors are reduced, resulting in consistently high-quality outputs. Staff can spend time on other field tasks and quality checks without worrying about waiting for data processing.

• Real-time team sharing: Because the data is in the cloud, stakeholders can simultaneously view and use the generated point cloud models. 3D data can be shared in real time between the field, the office, and even remote team members, reducing communication loss. With everyone referencing a single, up-to-date dataset, collaborative work can proceed without misunderstandings.

• Access to the latest technology at all times: Since the software is continuously updated on the cloud side, users can take advantage of the latest point cloud generation algorithms and AI analysis features without actively managing updates. There is no need to update or replace software in-house, and this ensures consistently high-precision results using cutting-edge technology.

• Contributes to surveying DX and labor reduction: These effects promote the digital transformation (DX) of surveying operations as a whole. Even with limited personnel, a wide area of sites can be covered and repetitive periodic surveys become easier, contributing to workstyle reforms and alleviating labor shortages. It also aligns with the Ministry of Land, Infrastructure, Transport and Tourism's promotion of *i-Construction*, and can be said to be a cutting-edge method that contributes to productivity improvement.

Use Cases of Cloud-based Automated Point Cloud Generation

What specific uses become possible on-site when point cloud generation in the cloud is realized? Here are some examples.

• Earthwork volume calculation and as-built management: Point cloud data is highly effective for calculating the volume of piled soil and excavation quantities at excavation and earthwork sites. By photographing the entire site from the air with a drone and processing the images into point clouds in the cloud, soil volumes for large sites can be computed in a short time. This is significantly faster than traditional manual cross-section surveying, enabling safe and highly accurate as-built management.

• Visualization and sharing of construction progress: If you regularly photograph the site and generate point cloud models, you can "visualize" construction progress in 3D. By overlaying the latest point cloud and design data in the cloud and comparing them, you can instantly identify over- or under-excavation and filling, as well as the installation status of structures. Sharing that information online with clients and stakeholders allows timely verification of construction status without visiting the site, facilitating smoother meetings.

• Investigation and surveying of hazardous areas: Even on steep slopes or disaster sites where it is dangerous for people to enter, surveying can be conducted safely through photography and cloud processing. For example, by photographing a landslide area from a distance with a telephoto lens and converting it into a point cloud, you can obtain detailed terrain information without approaching the cliff edge. From the resulting 3D model, it is also possible to measure required dimensions and slopes or to analyze areas of deformation.

Beyond these examples, the applications of cloud-based automated point cloud generation are wide-ranging, including bridge and tunnel maintenance, forest resource surveys, and urban area condition surveys. Once captured and digitized, point cloud data can be used to extract additional measurements later or to perform analyses for different purposes, which is another advantage of using point cloud data. By combining the optimal capture methods for each site (drone aerial photography, ground-based photography, mobile LiDAR scanning, etc.), it is possible to streamline and advance tasks that were previously difficult.

Key points to keep in mind when introducing

Cloud-based automatic point cloud generation is convenient, but there are points you should keep in mind to maximize its effectiveness. First, ensuring capture quality is important. In photogrammetry, blurred images or insufficient coverage will lead to reduced accuracy. Make sure to photograph the subject with sufficient overlap and to provide high-resolution, sharp photos. Also, when photographing large sites with a drone, it's essential to create an appropriate flight plan to eliminate blind spots.

Next, also consider network conditions and data volume. Uploading hundreds of high-resolution photos can take a long time. If on-site network conditions are poor, you should save the captured data to a PC or similar device and upload it in bulk after returning to the office. Even if cloud processing itself is fast thanks to parallelization, keep in mind that upload time is affected by the speed of your internet connection.

Additionally, utilizing positioning information is a key point. For surveys that require absolute coordinates, obtaining geotagged data in advance—by installing known control points or using drone RTK positioning, for example—makes aligning models in the cloud smoother. If you perform accuracy checks with ground validation points as needed, you can confidently use point clouds generated in the cloud as deliverables.

Finally, be sure to verify the cloud service's security measures that you will use. Since you are entrusting business-critical data, it is important to select a service that can be operated securely, including encrypted communications, access controls, and data backup systems. With a trusted platform, you can preserve and share data more reliably and efficiently than by managing it on-premises.

Simplified surveying with LRTK

One example of a service that realizes the "just shoot and send to the cloud" point cloud generation introduced so far is LRTK. LRTK (Eru Aaru Tī Kē) is a solution that leverages smartphones and the cloud to make 3D surveying easy for anyone. By simply launching the dedicated smartphone app and scanning and photographing the site, the acquired data is synchronized directly with the cloud. Without having to worry about cumbersome file transfers or PC processing, point cloud models are automatically generated and stored in the cloud.

On the LRTK Cloud, generated point clouds can be quickly displayed in a browser and inspected by freely changing viewpoints. Because 3D data can be shared with stakeholders on the web without installing special software, communication between the field and the office becomes smoother. LRTK also supports importing photos captured by drones and other surveying data, enabling the construction of high-accuracy 3D models that integrate multiple data sources. For example, by processing wide-area point clouds from drone aerial photography together with detailed point clouds from smartphone LiDAR, you can obtain a model that combines both broad scale and fine detail.

By leveraging LRTK in this way, surveying tasks that previously required hiring specialized contractors or expensive equipment can be carried out safely and quickly by your own staff alone. Because the entire process — from acquiring point cloud data to sharing it — is completed in the cloud, it also helps promote on-site DX (digital transformation). Companies that have actually introduced LRTK report comments such as "It was intuitive even for first-time users" and "We were able to grasp site conditions in 3D in a short time, which streamlined operations." With simplified surveying using LRTK that incorporates the latest technologies, please experience the productivity improvements in your surveying operations for yourself.

FAQ

Q. What preparations or equipment are required to perform point cloud generation in the cloud? A. In principle, you can get started with imaging equipment such as a camera-equipped smartphone, a digital camera, or a drone, together with an internet connection. Dedicated expensive laser scanners or high-performance PCs are not necessarily required. If you register to use a cloud service to upload the captured data, you can begin capturing and uploading data on site right away.

Q. Are the accuracies of point clouds created by photogrammetry inferior to those from laser scanner surveys? A. Point cloud data generated by photogrammetry can achieve high accuracy under appropriate conditions. With the latest algorithms, it is possible to obtain point clouds that fall within an error range on the order of a few centimeters (a few in), and for many civil engineering surveying applications they are sufficiently practical. However, because laser scanners measure distance directly, they can be superior in terms of accuracy and consistency. Ideally, you should choose based on the required level of accuracy, but nowadays there are increasing cases where photogrammetry + cloud processing produces results comparable to ground surveying.

Q. Can cloud services be used at sites with unstable internet? A. Online connectivity is best, but if the network is unstable you can handle it by uploading later. On site, first capture the photo data and temporarily save it on your PC or device. Move to an area with good reception and then upload everything to the cloud at once. Recently, portable Wi‑Fi and on-site relay devices have become more widely available, and there are increasing cases where mobile communication allows immediate cloud synchronization even in mountainous areas.

Q. Is it safe to store highly confidential surveying data in the cloud? A. If you use a reliable cloud service, security is carefully considered. Communications are encrypted, and strict security measures are implemented on the server side. In addition, access-rights settings prevent data from leaking to third parties. Rather than managing data solely on a local PC, backups are performed reliably, giving the advantage that data will not be lost even in the event of a disaster. If you choose a service that aligns with your company’s policies, you can be confident in the handling of confidential data.

Q. If I request point cloud generation in the cloud, how long does it take to receive the results? A. Processing time varies depending on the amount of data, but results are often available within a few hours after capture. For example, a point cloud model for a dataset of a few hundred photos is typically completed within several hours to about half a day. Processing that used to take a whole day or more on a conventional PC is automatically executed in the cloud during that time, allowing users to devote the waiting period to other tasks. Even with large volumes of data, if you upload them overnight you can often check the processing results the next morning, enabling more efficient use of time.