SfM processing × LRTK to streamline construction sites: Major productivity gains with smartphone GNSS

What is SfM? Photogrammetry and point cloud generation spreading in the construction industry

SfM (Structure from Motion) is a technology that automatically generates three-dimensional models (point cloud data and 3D meshes) from multiple photographic images. By analyzing many photos taken with drones, digital cameras, or smartphones, it matches common points to estimate camera positions and object shapes. This allows the construction of 3D site models without special surveying equipment, making it an increasingly popular method of photogrammetry in recent years.

In the construction industry, driven by initiatives like the Ministry of Land, Infrastructure, Transport and Tourism's i-Construction, 3D surveying using SfM has rapidly spread. High-precision point clouds and orthophotos can be created quickly from drone imagery or smartphone photos, contributing to improved operational efficiency and cost reduction. What used to require specialized contractors can now be trialed easily with handheld camera equipment, and SfM processing is becoming standard as a DX tool for civil engineering and construction sites.

From as-built management to volume calculation: Use cases of SfM on construction sites

Point cloud data and 3D models generated by SfM are used across many construction site tasks. Major application scenarios include:

• As-built management (quality verification): Record post-construction terrain and structures with SfM point clouds and compare them to design data to verify dimensional conformity. Discrepancies down to millimeters can be detected, making it useful for checking whether embankments and slopes match design specifications. Photo-based 3D models also include color information, aiding visual inspection of finish quality.

• Progress management: By photographing the site at each stage and creating SfM models, you can understand construction progress in three dimensions. For example, weekly drone surveys that are converted into point clouds allow remote confirmation of how much excavation or embankment work has been completed. Details that are hard to discern on plan drawings become obvious in 3D, making it easier to share progress between the field and the office.

• Surveying and current-condition assessment: SfM can survey wide areas in a short time, even for complex terrains that tend to be missed by traditional total stations or GPS surveys. Current terrain of forests or development sites and locations of cracks in structures can be modeled in detail from photos, allowing arbitrary dimensions and cross-sections to be derived later. Large-scale point cloud capture records vast numbers of points at once, so it typically has fewer omissions than manual surveying.

• Earthwork volume calculation: SfM point clouds make volume calculations for embankment and excavation more efficient. For example, comparing pre- and post-construction terrain models to compute volume differences or measuring stockpile volumes via photogrammetry. Whereas cross-section surveys and analysis used to take time onsite, SfM models can automatically compute volumes for the needed area, enabling immediate grasp of earthwork quantities.

In this way, photogrammetry using SfM is being applied practically across site activities from quality control to progress monitoring and quantity estimation. The revolutionary aspect is that necessary 3D information can be obtained from photos taken with consumer cameras or smartphones, leading to increased adoption even on small- to medium-scale construction sites.

Conventional photogrammetry challenges: GCP placement and the effort of georeferencing

To align point clouds and models produced by SfM with real-world coordinate systems and ensure high survey accuracy, conventional approaches have faced certain challenges. A representative challenge is the placement of GCP (Ground Control Point) targets.

In photogrammetry, it is common to place multiple ground markers with known coordinates (GCPs) and use those coordinates in processing to correct the generated 3D model to the correct scale and position. For instance, when aerially surveying a large development site, you might set about 5–10 GCPs at corners and near the center, ensuring they appear in the images. In software, you then mark the GCP locations in each photo and assign their known coordinate values as constraints to align the model to the geodetic coordinate system.

However, GCP placement requires effort and manpower. Each point needs high-precision coordinate measurement with surveying instruments, and the positions must be checked during photo analysis—this becomes a large burden for vast sites or frequent surveys. When insufficient GCPs are placed, SfM results may remain in a local relative coordinate system, leading to scale errors and positional offsets relative to real coordinates. In addition, at sites with poor access or dangerous areas, it may be physically impossible to place markers, making the routine of placing GCPs every time impractical.

Thus, "how to provide accurate real-world coordinates to photogrammetric models" has been a traditional challenge, but solutions have emerged in recent years. One such solution is direct georeferencing via RTK-GNSS. In particular, technologies like LRTK used with smartphones are enabling high-precision alignment without relying on GCPs.

High-precision positioning and automatic geotagging of photos with smartphone GNSS (LRTK)

LRTK (Local RTK) uses small RTK-GNSS receivers attachable to smartphones to achieve centimeter-level positioning accuracy. Typical built-in smartphone GPS accuracy is around 5–10 meters, but by using an LRTK device and a dedicated app, correction data (RTK) is applied to satellite positioning, enabling current positions to be determined with errors within a few centimeters. For example, correction data can be obtained from reference stations over the internet, or centimeter-level augmentation signals (CLAS) from Japan’s quasi-zenith satellite "Michibiki" can be received, dramatically reducing smartphone positioning errors.

With such high-precision GNSS, it becomes possible to automatically attach accurate geotags (position coordinates) to every photo taken. While standard smartphone photos include latitude and longitude tags, their significant errors make them insufficient for 3D reconstruction. With LRTK, the phone’s position at each shutter moment can be recorded to centimeter accuracy, turning each photo into a survey observation point. During SfM processing, these photos with high-precision coordinates are used for initial matching, so the entire model is automatically aligned to the site coordinate system.

Specifically, when photos are taken with a smartphone fitted with an LRTK device, a dedicated app records position information (latitude, longitude, altitude) with real-time corrections. The position data are uploaded to the cloud along with the photos, and during SfM analysis, the photographic positions are treated as known points in bundle adjustment. As a result, without placing GCPs, the resulting point clouds and models are already correctly scaled and oriented in global coordinates, eliminating the need for time-consuming coordinate transformations or post-processing and dramatically speeding up the time from shooting to obtaining results.

In addition, camera attitude can be estimated from smartphone angle sensors (IMU) and image feature detection, enabling high-precision photogrammetry with minimal required information. The combination of accuracy correction via LRTK and automatic geotagging is a groundbreaking mechanism that significantly lowers the barrier to ensuring accuracy in field photogrammetry.

Workflow for smartphone-based site surveying and labor-saving effects (single-person operation, blind-spot coverage)

A major advantage of smartphone + LRTK site measurement is that anyone can easily perform it alone. Below is the basic workflow and the labor-saving effects.

• Site photography: An operator attaches an LRTK receiver to a smartphone and walks around the area to be surveyed while taking photos. While drone imagery can be combined for high or wide areas, handheld smartphone photography can reliably capture confined spaces and the backsides of structures that drones cannot see, ensuring comprehensive data capture.

• Automatic recording and tagging: Each time the shutter is pressed, the smartphone app determines the current position and embeds a high-precision location tag into the photo file. Positioning is done with a single button, so accurate coordinate capture is possible without special surveying skills. Observations that traditionally required two people (one holding a prism and another operating the survey instrument) can be replaced by a single smartphone, contributing to reduced staffing.

• Data upload: After shooting, photos are uploaded directly from the smartphone to a cloud service. There is no need to bring data back to the office for PC processing—the data transfer can be completed from the field. This too is a single-person workflow.

• SfM processing and point cloud generation: SfM processing is automatically executed in the cloud to generate point clouds and 3D models (offloading processing to the cloud rather than the smartphone makes it easy). Thanks to LRTK’s high-precision geotags, the resulting models already align accurately to map coordinates, eliminating additional alignment work.

• Review and utilization of results: Generated point clouds and models in the cloud can be viewed immediately from the site via smartphone. You can measure distances, areas, and volumes on the spot, or cross-check with record photos. This removes the need for post-processing in office PC software, allowing real-time use of results on site—a key labor-saving point.

Through this sequence, site managers themselves can complete current-condition measurements in a short time without assembling specialized surveying teams. In an industry facing severe labor shortages, smartphone surveying that enables single-person operation is a key method for labor reduction and efficiency gains. Moreover, locations that were previously difficult to measure (under viaducts, behind slopes, etc.) can be supplemented by close-range photography, achieving comprehensive data capture and improving the reliability of quality control.

Remote sharing and real-time utilization enabled by SfM × LRTK × cloud integration

Combining smartphone SfM, LRTK, and the cloud enables instant sharing of measured data and real-time utilization. Uploading field-acquired information directly to the cloud promotes data use beyond geographic limits.

For example, once a field operator uploads point cloud measurements from a smartphone to the cloud, office engineers or clients can view the 3D point cloud in a web browser immediately. No dedicated software installation is required, and the latest site conditions can be checked anytime, anywhere via the internet. This makes it easier to share construction status remotely with supervisors or partner companies, facilitating faster decision-making.

Cloud point cloud viewers also allow measurements such as distances between any two points, areas, and volumes, enabling data-driven analysis without visiting the site. For instance, head office engineers could check daily updated excavation point clouds and determine as-built conformity or quantity verification on the same day. Time spent on onsite checking and reporting is reduced, realizing data-driven remote construction management.

Another advantage of cloud integration is the ability to overlay design data. If design 3D models or drawings are registered in the cloud in advance, uploaded as-built point clouds can be automatically aligned, allowing on-the-spot visualization of differences between design and construction. This makes it possible to instantly grasp, for example, the difference between planned and actual earthwork volumes or misalignment of installed structures, helping to prevent rework. Tasks that previously required designers to visit the site can now be confirmed in the cloud, reducing travel time and enabling remote inspections.

Thus, SfM × LRTK × cloud integration not only streamlines data capture but also optimizes downstream utilization processes. Immediate sharing of 3D data among stakeholders and rapid analysis and decision-making contribute to accelerating and optimizing the entire construction PDCA cycle.

Lowering the adoption barrier and operational flexibility with smartphone use (suitable for SMEs and local governments)

Adopting new technology often requires significant investment and specialized personnel, but smartphone × GNSS surveying has the advantage of a low adoption barrier. By leveraging smartphones that people are already familiar with, 3D measurement can begin on site without special equipment or extensive training.

From an equipment perspective, all you need is a smartphone and a small GNSS receiver, so initial costs are lower compared to 3D laser scanners or high-end RTK surveying instruments. Avoiding expensive dedicated equipment makes it easier for small contractors and regional builders to adopt ICT construction methods, letting them widely enjoy the benefits of ICT construction. If existing smartphones can be repurposed, the burden is reduced further.

Operationally, intuitive smartphone apps make learning easy, so site supervisors or staff can perform surveying without being dedicated surveyors. For example, municipal staff can use smartphone SfM for road maintenance inspections, or local government personnel can quickly measure steep terrain in disaster-affected areas. In fact, cases of local governments and SMEs using smartphone GNSS are already emerging, aiding disaster-site situational awareness and simple land surveys.

Smartphone-based measurement also offers scalability and flexibility. It is easy to combine with drones, 360° cameras, or smartphone-integrated LiDAR scanners as needed, enabling flexible data acquisition tailored to site conditions. For example, you might normally use handheld smartphone surveys, supplement with drone SfM for large-scale earthworks, and use the phone’s LiDAR for indoor equipment inspections—integrating multiple methods on a single platform. If such data are managed centrally in the cloud, smart site information management can be realized regardless of company size.

The ease and cost reduction of smartphone utilization can therefore be a driving force to spread digital surveying to small operators and municipalities that had been reluctant to adopt ICT. Broadening the technology’s reach is expected to raise industry-wide productivity.

Conclusion: The future of simplified surveying opened by smartphone GNSS (LRTK)

Combining smartphones with high-precision GNSS in LRTK technology is opening new horizons in the field of photogrammetry (SfM). What was once a high-barrier 3D survey can now be performed by anyone, quickly, and at low cost, transforming how construction sites capture and use data.

Smartphone SfM that yields accurate point clouds and models without GCP setup is likely to become an increasingly standard tool on sites. Especially amid labor shortages and work-style reforms, single-person rapid surveying methods are a key to labor saving and productivity improvement. Real-time cloud-shared data also enable a new construction management style where sites can be monitored and guided remotely.

With further improvements in smartphone GNSS accuracy and the spread of 5G communications, the era in which a smartphone itself becomes an advanced surveying instrument is poised to arrive. This will allow sites and small operators previously distant from ICT construction to benefit from digitalization, raising efficiency and quality across the industry. The shift from “surveying being the domain of specialists” to “surveying becoming part of routine tasks anyone on site can perform” is imminent—smartphone GNSS (LRTK) is bringing that simplified surveying future within reach. Keep an eye on how this field-driven innovation continues to spread.