Photogrammetry vs 3D Laser Scanning: Comparing Accuracy and Characteristics of 3D Point Cloud Measurement with Smartphone RTK

In recent years, the use of three-dimensional point cloud data has been rapidly advancing in surveying and CIM/BIM fields. Representative acquisition methods are photogrammetry, which generates 3D models from photographs, and 3D laser scanning, which acquires point clouds by directly measuring distances with laser light. This article focuses particularly on comparing their measurement accuracy and work time, delving into the technical characteristics, use cases, strengths, and weaknesses of each. It also touches on new point cloud acquisition methods using smartphones and RTK (real-time kinematic), such as LRTK, and considers future prospects.

What is photogrammetry

Photogrammetry is a technique that reconstructs the three-dimensional shape of an object from multiple photographic images. The subject is photographed from various angles using drones, single-lens cameras, or, recently, smartphones, and the camera positions and object shape are estimated by matching feature points between images in software. The resulting point clouds and 3D models include color image textures, so they are characterized by their photorealism and visual clarity. Equipment can be ordinary cameras, making it relatively low-cost to introduce.

However, achieving high-precision measurements with photogrammetry requires several conditions. First, the photos taken must have sufficient resolution and overlap. Shoot so that object details are clearly captured and each part appears in multiple photos. It is also important that the subject has sufficient texture (patterns or features). Subjects that lack detectable features, such as glass, water surfaces, or plain white walls, or shooting in dark or nighttime conditions, will reduce accuracy or make reconstruction difficult. Furthermore, to improve the absolute coordinate accuracy of the obtained point cloud, it is necessary to take measures such as placing known-coordinate markers called control points on site or using high-precision GPS during shooting. If these are done appropriately, photogrammetry can yield survey results with cm-level accuracy (half-inch accuracy).

What is 3D laser scanning

3D laser scanning is a method that irradiates laser light onto an object and captures the reflected light with a sensor to directly measure distances and create point clouds. The commonly used devices are called 3D laser scanners, with various types such as terrestrial units (TLS: Terrestrial Laser Scanner) that are tripod-mounted and scan wide areas while rotating, mobile mapping units mounted on vehicles or drones for measurements while moving, and handheld scanners. Laser scanners can capture hundreds of thousands to millions of points per second, and their strength is obtaining high-density point clouds of complex structures or terrain in a short time.

Point clouds obtained by laser scanning are geometric collections of points measured directly by distance, and some devices can also provide reflectance intensity or color information from RGB cameras. Unlike photogrammetry, laser scanning can measure in dark conditions and is less affected by object material or lighting conditions. However, transparent or reflective materials such as glass or mirrors, or very black surfaces (materials that absorb a lot of light), may not return laser signals and can be difficult to measure. Dedicated equipment is also expensive (millions of yen and up), large, and tends to require specialized knowledge to operate. Recently, some high-end smartphones have been equipped with small LiDAR sensors, enabling simple short-range 3D scanning, so laser scanning technology is also becoming more portable.

Comparison of measurement accuracy: how accurate are the point clouds

Measurement accuracy is one of the most important points in comparing photogrammetry and laser scanning. Generally, laser scanning is considered more accurate, but the situation varies depending on conditions and methods.

• Photogrammetry accuracy: In photogrammetry, reconstruction accuracy is influenced by photo resolution, camera calibration, and algorithm accuracy. With proper shooting and processing, it is possible to obtain point clouds for architectural or terrain surveys with an error on the order of a few centimeters (cm-level accuracy, half-inch accuracy). For small objects shot at close range, millimeter-level accuracy (≈0.04 in) can also be achieved. However, photogrammetry inherently varies in accuracy depending on camera placement and shooting angle relative to the subject, and vertical accuracy tends to be worse than planimetric accuracy. In aerial photogrammetry, which mainly uses overhead shooting, vertical errors can be two to three times those of horizontal position. As mentioned earlier, ensuring absolute positional accuracy depends on control points or geotagging with high-precision GPS. If a model is created from photos alone without these, the entire model can float in an arbitrary coordinate system, causing scale or position discrepancies with reality.

• 3D laser scanning accuracy: Because laser scanners measure distance directly in principle, the ranging accuracy at each point is very high. High-performance terrestrial laser scanners exist with stated accuracies of several millimeters (several 0.04 in), and they are used in applications requiring precision, such as displacement measurement of medium-sized structures or dimensional measurement of plant piping. However, laser scan accuracy also depends on distance, and at long ranges, laser beam divergence and small reflection deviations increase errors (e.g., at 10 m (32.8 ft) distance, errors on the order of ± a few millimeters (± a few 0.1 in) may occur). On the other hand, equipping scanners with high-precision attitude and positioning sensors can secure absolute coordinate accuracy of the acquired point cloud. For example, mobile mapping systems with high-precision GNSS and IMU can align broad-area surveys to map coordinates within a few centimeters (a few in). Because laser scanning has less bias due to shooting angle, it provides consistently high accuracy including vertical direction, which is a major difference from photogrammetry. Also, for surfaces under vegetation where photographs cannot see the ground, if some laser pulses reach the ground (using multiple echoes), point clouds can be obtained; therefore, in forest surveying, laser often outperforms in terrain representation accuracy.

In summary, photogrammetry can be expected to provide high accuracy when the environment is open and well-lit and the subject can be approached closely, but laser scanning is more reliable for complex terrain, under vegetation, or nighttime measurements. In practice, the strengths of each are often combined to complement one another as needed.

Comparison of work time: how efficient are data acquisition and processing?

Another important comparison axis is the difference in work time and effort required from data acquisition to deliverables. Photogrammetry and laser scanning have significantly different workflows, so it is not always clear which is faster, but the following are typical tendencies.

• Photogrammetry workflow and time: On-site work for photographing is often relatively short. For example, aerial photography with a drone can cover wide areas in a short time (tens of minutes) using an automatic flight program, and ground photography is mobile because a person can simply walk around with a camera. The ability to perform tasks with a small team is a strength of photogrammetry. However, note that post-processing can take time. Generating point clouds or 3D models from many photos requires several hours to, in some cases, tens of hours of processing on a PC (structure analysis and point cloud synthesis). Depending on data volume and PC performance, it is not uncommon for processing to take more than one night. Also, achieving high accuracy requires time and effort on-site for placing and surveying control points. Thus, the typical pattern is “field work is fast, but analysis takes time.”

• 3D laser scanning workflow and time: Laser scanning itself on-site takes a certain amount of time. For fixed terrestrial scanners, a single station (setup location) typically requires a scan of several minutes to ten-odd minutes, and this must be repeated to cover the entire area. Large facilities may require repositioning dozens of times, so on-site measurement time tends to be longer than photogrammetry. On the other hand, the acquired point cloud can be confirmed on site immediately, so areas needing re-shooting can be recognized in real time. Post-processing involves point cloud registration (alignment) between multiple stations and noise removal, but automation in dedicated software has advanced, and processing time is often shorter than image analysis, depending on data volume. Also, with the spread of mobile mapping and handheld laser scanners, continuous measurements while walking and later point cloud generation by SLAM algorithms eliminate the need for fixed-station registration. In such mobile measurements, point cloud construction can progress in parallel with fieldwork, enabling rapid coverage of wide areas.

In summary, photogrammetry features speedy field shooting but lengthy post-processing, whereas laser scanning requires more on-site measurement time but produces data that is readily usable immediately. Which is more efficient depends on project scale and delivery requirements.

Other characteristics and appropriate use cases

Beyond accuracy and work time, photogrammetry and 3D laser scanning differ in various ways. Understanding these differences and choosing appropriately for the application is important. The main points are summarized below.

• Equipment cost and ease of operation: Photogrammetry can usually start with relatively inexpensive equipment such as cameras and drones, and shooting can be done by one person. In contrast, high-precision laser scanners have high initial costs and require expertise for setup and operation. For example, for daily progress recording on a construction site, taking photos with a handheld camera or smartphone and modeling them is more convenient and economical. Conversely, when precise deformation measurement of important structures is required, it may be worth deploying expensive equipment like laser scanners.

• Expressiveness of data: Models and point clouds from photogrammetry include color information and reproduce appearance faithfully, making them intuitive to understand. For instance, in CIM use to convey site conditions to stakeholders, photo-based 3D models are persuasive. Laser scanning point clouds primarily contain geometric information, and surface color or texture is often limited to simple intensity values (recent models, however, combine scanners with high-resolution cameras to obtain color point clouds). To make laser-derived point clouds more visually understandable, rendering software is used for false-coloring, or photogrammetry-derived photos are combined for texture mapping.

• Environmental adaptability: As mentioned, photogrammetry struggles when subjects lack patterns or in dark environments. Conversely, laser scanning can measure in tunnels or at night without problems. While aerial photogrammetry is powerful for wide-area terrain surveys from above, photographing from above alone is insufficient to capture ground beneath forests or grass. In that case, airborne LiDAR (drone or aircraft-mounted) that directly captures the ground is effective. Conversely, for structural inspections like concrete crack surveys, high-resolution photo-based photogrammetry models can reveal fine texture details. Thus, suitability depends on the subject and environment, and it is important to “use the right tool for the purpose,” not simply which is superior.

Based on the above, typical use cases include:

• Earthwork volume measurement and progress management on construction sites: In open development sites, drone photography and photogrammetry make it easy to measure terrain and earthwork volumes, and frequent captures (e.g., weekly) for point cloud comparison are feasible. It covers wide areas quickly and is economical. However, before forest clearing or for immediate nighttime post-work measurement, laser scanners or drone LiDAR are advantageous.

• As-built modeling of buildings and plants (inventory modeling): High-precision laser scanning is suitable for measuring complex factory piping or building interiors to millimeter accuracy for dimensional surveys and BIM modeling. Conversely, for digital archiving of building exteriors or ruins, photogrammetry can create 3D models including texture for stakeholder sharing or VR use. Combining both is also used: laser scanning for geometry and high-resolution photos for texture mapping.

• Infrastructure inspection and maintenance: High-precision laser scanning is indispensable for quantitative measurements of tunnel or bridge displacement and deflection. Photo-based models are useful for recording concrete surface deterioration or exposed rebar. For road or trackside patrol inspections, attempts to reconstruct 3D from camera images (like dashcam footage) are increasing, advancing photogrammetry use. For wide-area transmission line surveys, helicopter- or drone-mounted laser scanners are used, showing that choice depends on the object in infrastructure management.

New point cloud acquisition methods using smartphone RTK

A notable recent topic is point cloud measurement combining smartphones with RTK GNSS. For example, systems called LRTK equip a smartphone with a high-precision RTK receiver to easily obtain 3D point clouds by photo capture or LiDAR scanning. RTK (real-time kinematic) is a technique that reduces satellite positioning errors in real time with correction information from a base station, achieving centimeter-level positioning accuracy (cm-level accuracy, half-inch accuracy). Traditionally, achieving high absolute accuracy in photogrammetry required placing many control points on the ground, but with smartphone RTK, shooting positions themselves are recorded with cm-level accuracy, allowing omission of cumbersome control work. By combining smartphone internal accelerometers and gyros, shooting attitude and orientation can also be recorded, directly attaching geographic coordinates to the acquired point cloud.

The advantages of smartphone RTK methods are their ease and mobility. Without carrying dedicated large equipment, measurement can be completed simply by holding a smartphone on site. For example, with an LRTK system, using smartphone LiDAR for close-range scanning can obtain point clouds on the spot and immediately calculate earthwork volumes (effective up to about 8 m (26.2 ft)). If you want to record a wider area at higher detail, you can switch to taking multiple photos with the smartphone and processing photogrammetry in the cloud, flexibly using the strengths of LiDAR and photogrammetry. Tasks that were difficult with traditional laser scanners, such as bringing equipment into confined spaces or measuring at height, become easy with a smartphone because people can move and shoot. The obtained point cloud is positioned in geodetic coordinates (latitude/longitude, etc.) from the start, making it easy to overlay with other GIS/CIM data.

Of course, smartphone sensors are limited compared to dedicated equipment and cannot replace all high-precision control surveys or large-scale topographic mapping. However, for field checks, simple surveys, intermediate construction checks, and rapid disaster site recording—situations where speed and reasonable accuracy are required—they are highly effective. Municipalities that have introduced smartphone RTK systems like LRTK have reported cases where staff quickly measured and point-clouded post-disaster damage themselves and used the data in recovery planning. As smartphone sensor technology advances, such easy measurement methods may become one of the surveying standards.

Conclusion and future prospects

Photogrammetry and 3D laser scanning are three-dimensional measurement technologies with different strengths. Photogrammetry has lower cost and introduction barriers and easily produces visually appealing models, but attention is needed regarding accuracy and environmental conditions. 3D laser scanning offers high and stable point cloud acquisition and performs well under difficult conditions, but there are hurdles in equipment cost and operation. As compared in this article, both measurement accuracy and work time have trade-offs, and choosing between them depends on site conditions and objectives.

Going forward, fusion of the two technologies and further miniaturization and cost reduction are likely. Examples include proliferation of small laser scanners mountable on drones and algorithm improvements in photogrammetry software to reduce processing time. Moreover, the emergence of new methods using smartphone RTK (such as LRTK) is accelerating the trend toward “easy high-precision 3D measurement for anyone.” An era is approaching in which field personnel can quickly acquire point cloud data themselves and feed that data back into CIM/BIM and maintenance management.

By correctly understanding the characteristics of photogrammetry and laser scanning and selecting the optimal method for the objective while actively adopting new measurement technologies, surveying and point cloud utilization on site will become even more efficient. Consider exploring forms of 3D point cloud utilization that fit your company’s operations, including the latest solutions such as simple surveys using smartphone RTK.