How far can markerless 3D measurement go? Five tips to improve accuracy

In recent years, the construction and civil engineering industries have been increasingly adopting 3D measurement techniques using drones, laser scanners, and photogrammetry. 3D measurement can improve efficiency and sophistication across a variety of tasks—from structural inspections to earthwork volume calculations, quality control, slope checks, and documenting site development.

Traditionally, to ensure measurement accuracy, markers (control points) were placed on the ground and used as references for aligning and correcting data.

However, placing and surveying markers is labor-intensive and can be hazardous in some locations. For that reason, demand has grown for efficient 3D measurement methods that do not require marker placement.

So, how feasible is markerless 3D measurement in practice?

This article answers that question by explaining the latest technology trends and key points for performing markerless 3D measurement. Focusing on five tips to improve accuracy, it provides concrete guidance on using drones, tripod-mounted LiDAR, photogrammetry, smartphone measurement, and other methods.

Table of contents

• How far is markerless 3D measurement possible?

• Tip 1: Choose the appropriate 3D measurement method and equipment

• Tip 2: Thorough preparatory work tailored to the environment

• Tip 3: Ensure sufficient overlap and multi-angle measurement

• Tip 4: Use positioning technologies such as GNSS and IMU

• Tip 5: Careful post-processing and accuracy verification

• Conclusion

How far is markerless 3D measurement possible?

With the latest 3D measurement technologies, there are an increasing number of practical cases where markerless measurement can achieve sufficient accuracy. For example, photogrammetry using a drone equipped with RTK-GNSS can sometimes generate orthophotos and point clouds with accuracy on the order of a few centimeters (a few inches) without establishing ground control points (GCPs). Ground-based 3D laser scanners are also seeing techniques that stitch scans from multiple positions using automatic feature matching, enabling wide-area measurement without placing targets (reflective stickers, etc.). Handheld (mobile) 3D scanners using SLAM technology allow an operator to walk while the device estimates its own position and captures the surrounding environment, removing the need for complex target placement. Recently, apps have also appeared that enable simple 3D scanning or photogrammetry using LiDAR sensors or cameras built into smartphones and tablets.

That said, whether markerless measurement is appropriate depends on the required accuracy. For tasks demanding strict centimeter-level accuracy (centimeter-level accuracy (half-inch accuracy)), such as final quality control, a completely markerless approach may produce errors outside acceptable tolerances.

When measuring without markers, challenges can remain in aligning to a surveying coordinate system (absolute accuracy) and in correctly scaling the model. However, by combining high-precision GNSS, IMUs, and other technologies, these challenges are being gradually overcome.

In short, it is important to assess the required accuracy for the intended use and choose the appropriate method and equipment to meet that level. Next, we introduce five concrete tips to improve accuracy when performing markerless 3D measurement.

Tip 1: Choose the appropriate 3D measurement method and equipment

To achieve high-accuracy markerless 3D measurement, selecting the right measurement method and equipment is crucial. 3D measurement options include drone photogrammetry, tripod-mounted laser scanners, mobile scanners, and smartphone measurement; each has different strengths and accuracy characteristics. Choosing the optimal method according to the scale of the target, the environment, and the required accuracy reduces unnecessary errors and increases the likelihood of success without markers.

For example, drone photogrammetry is efficient for surveying terrain and calculating earthwork volumes over large development sites because data can be acquired from the air in a single pass. Since the simple GPS built into many off-the-shelf drones typically yields position accuracy on the order of tens of centimeters, it is advisable to use drones equipped with high-precision RTK-GNSS from the outset or to apply PPK (Post-Processed Kinematic) correction in post-processing. With an RTK-capable drone, accurate coordinate tags are attached to captured images, dramatically improving the positional accuracy of the resulting 3D model without installing numerous ground control points.

For detailed shape measurement of structures and quality control, tripod-mounted 3D laser scanners can provide millimeter-level accuracy (approximately 0.04 in). However, covering wide areas with tripod-mounted units requires merging scans from multiple setups, which typically involves placing several targets (spheres or sticker-type markers) for alignment. When stitching scans without markers, ensure substantial overlap between scans and include natural features such as building corners and terrain characteristics to facilitate feature-based matching during post-processing. Moving the scanner incrementally over short distances to limit accumulated error is also effective.

Handheld (mobile) laser scanners, increasingly common today, use SLAM to perform real-time self-localization while walking and can capture large point clouds quickly. They are advantageous in places where moving tripod-mounted equipment would be time-consuming—such as inside tunnels or plant facilities. Although their absolute accuracy is generally inferior to tripod-mounted units (typically on the order of a few centimeters (a few inches)), their ability to autonomously align data in environments where target placement is difficult is a major benefit.

Smartphone- and tablet-based 3D measurement has also emerged. For example, modern smartphones often include simple LiDAR or high-performance cameras, and dedicated apps can scan surroundings to generate point clouds or mesh models. Smartphone measurement is highly mobile and convenient but offers limited accuracy compared to professional equipment—errors of a few centimeters to 10 cm (a few inches to 3.9 in) can occur. Thus, while useful for rough site records or early-stage design, smartphones are not ideal when precise quality control or design verification is required.

Selecting the method that best matches the target and purpose is the first step to successful markerless measurement. With appropriate equipment choices, you can often omit on-site marker placement while still meeting accuracy requirements.

Tip 2: Thorough preparatory work tailored to the environment

In markerless measurement, the environmental conditions at the time of capture and the state of the target directly affect data quality. Thorough preparatory work to create conditions optimal for measurement is the key to ensuring accuracy. Paying attention to weather, lighting, and the surface condition of targets can significantly reduce noise and errors.

First, choose suitable weather and lighting conditions for outdoor measurements. Do not fly drones or perform laser scans in rain or strong winds. Besides the risk of equipment getting wet, wind can cause drone instability or tripod shake, resulting in misaligned point clouds. Bright direct sunlight can create high-contrast shadows in photos and reduce matching accuracy in photogrammetry. If possible, aim for overcast days or shoot in the soft light of early morning or late afternoon (but be mindful of sunset times). Avoid long-distance measurements during heat haze in hot weather, as mirage near the ground can distort laser and photographic data.

For indoor or nighttime measurement, ensure sufficient lighting. Relying too heavily on high ISO in photogrammetry increases noise, so use portable lights to reduce dark areas and harsh shadows when needed. Laser scanners also benefit from adequate lighting because their integrated cameras may struggle to capture textures in dark conditions, making post-processing easier if lighting is sufficient. It is also important to eliminate moving objects from the measurement area. Heavy foot traffic or moving machinery can produce motion blur or ghosting in photos and cause point cloud gaps or duplicate captures, hindering accurate data merging. Where possible, restrict access during capture or temporarily pause work to obtain data in a static environment.

Pay attention to the "photographability" of the subject and surroundings. In photogrammetry, surfaces lacking patterns or features make it difficult for software to find keypoints for alignment. For uniformly white walls or pavement, temporarily adding tape or sticker patterns (not as survey markers but as visual textures), or using spray chalk to create removable random marks can help the software match images accurately. Be sure to use materials that can be removed afterward. Highly reflective surfaces such as metal or glass are difficult to capture accurately with lasers or cameras; consider applying anti-reflective spray or using a polarizing filter on the camera as appropriate, but take care not to damage the target surface.

Also, prepare and calibrate equipment before use. For drones, perform compass and IMU calibration before flight and verify the RTK base station or network RTK settings. For cameras, check lenses for dirt, set autofocus appropriately prior to shooting or lock to manual infinity focus, and perform test shots to check for overexposure or motion blur. For tripod-mounted scanners, confirm the instrument is level using a spirit level and run any available calibration mode before use. For handheld scanners, slowly look around the scene before starting to stabilize the SLAM initial map. For smartphone measurement, reset AR tracking once before starting and move the device slowly to let it scan the surroundings before beginning the main capture to improve accuracy.

By preparing the environment and ensuring equipment is in optimal condition, you can capture high-quality point clouds and images without markers, which in turn makes subsequent alignment smoother and leads to more accurate models.

Tip 3: Ensure sufficient overlap and multi-angle measurement

To integrate multiple data captures without markers, it is essential to ensure ample overlap between photos or scans and to capture the target from various angles. Insufficient overlap makes it difficult to correctly align adjacent photos or point clouds, causing distortions or misalignments in the model. Conversely, substantial overlap and multi-angle capture make it easier for software to detect common features, enabling high-accuracy data fusion even without markers.

In photogrammetry, prioritize high overlap. For drone terrain surveys, aim for at least 70–80% forward and side overlap between adjacent photos. For example, when shooting a flat area from the air, fly multiple parallel flight lines with approximately 80% lateral overlap between adjacent lines. For terrain with strong relief or structures with significant height differences, shoot from multiple altitudes and angles rather than keeping altitude constant. For buildings, supplement nadir (straight-down) photos with oblique images from all sides so walls are well captured. If each part of the target appears in at least three photos, feature matching in post-processing will be stable and produce point clouds with minimal misalignment.

Overlap is also important in laser scanning. When using a tripod-mounted laser scanner from multiple positions, ensure each scan position has overlapping fields of view with neighboring positions. Areas that are blind spots in one scan can be covered from another angle to prevent data gaps. For example, when scanning inside a tunnel, move the scanner at regular intervals and ensure each scan overlaps the adjacent segments. For exterior building surveys, plan positions to illuminate corners and curved surfaces from at least two directions so the entire structure is encompassed. This approach ensures common features (like wall corners or ground patterns) are captured across scans, facilitating smooth registration in post-processing. Avoid placing setups too far apart; expanding the coverage incrementally helps reduce accumulated error.

For mobile SLAM scanners, the path taken directly affects accuracy. A never-returning, single-pass route can allow sensor errors to accumulate unchecked, leading to drift and eventual positional error. Whenever possible, perform a loop closure by returning to the start point; overlapping scans at the departure point let the SLAM algorithm self-correct and reduce overall distortion. If a loop is impractical, pause at key locations to scan the surroundings thoroughly and capture enough features before proceeding. Moving at a consistent, slow pace is important—moving too quickly or jogging will produce coarse data and may cause tracking loss (tracking loss), so maintain steady slow motion.

Also, aim to capture the target from many angles. Features and hidden surfaces that cannot be captured from a single direction are faithfully reproduced only when measured from multiple viewpoints. For earthwork volume surveys, point cameras or lasers toward the toes of fills and deposits so slopes and typically shadowed lower areas are not missed. For structural inspections, capture surfaces with cracks from both head-on and oblique angles to produce 3D data that is easy to interpret later. By collecting multi-faceted data, each dataset can complement the others and enable high-accuracy 3D model construction without relying on markers.

Tip 4: Use positioning technologies such as GNSS and IMU

To obtain measurement data with correct positions and dimensions without markers, it is effective to leverage external positioning technologies whenever possible. Combining satellite positioning (GNSS) and inertial measurement units (IMU) can attach absolute position and orientation information to datasets, greatly improving overall model accuracy.

High-precision GNSS is a representative example. In drone photogrammetry, an RTK-GNSS-capable aircraft can tag every captured image with centimeter-level coordinates (centimeter-level coordinate tags (half-inch accuracy)). Conventional GPS can have errors of several meters (several ft), which can shift the entire model by that amount, but RTK provides high-precision positions for each photo so aerial triangulation converges to accurate scale and spatial relationships. Some tripod-mounted laser scanners can record the scan start location with an internal GNSS or be directly georeferenced by an external GNSS receiver, allowing scan data to be placed into geospatial coordinates. When surveying large sites over multiple days, embedding GNSS-referenced position information in each day’s data enables accurate integration later without control points.

Mobile SLAM scanners can also benefit from GNSS assistance in outdoor, GNSS-available environments. As SLAM inevitably accumulates small positional errors over long walks, obtaining high-precision GNSS positions at intervals can reset errors and allow scanning to continue with high accuracy. Some advanced mobile mapping systems fuse SLAM with real-time GNSS and IMU data to suppress drift and measure large spaces with centimeter-level accuracy (a few centimeters (a few inches)).

Recently, solutions have appeared to attach high-precision GNSS receivers to smartphones and tablets for photogrammetry or LiDAR scanning. Built-in smartphone GPS typically provides meter-level accuracy, but connecting an RTK-capable GNSS receiver via Bluetooth or a dedicated mount can enable near-survey-grade positioning on a phone. Even when taking ground photos, if each image has an accurate capture-location tag, subsequent image alignment is less likely to suffer scale or positional errors. Point clouds obtained with a handheld smartphone become easier to integrate into an absolute coordinate system if GNSS coordinates are attached.

Make effective use of IMUs and electronic compasses as well. Modern drones and 3D scanners include high-performance IMUs that sense device orientation and motion and embed that information in the data. Photo-processing software can use approximate camera orientations (Euler angles) and positions as initial values to accelerate algorithm convergence. Handheld scanners use IMUs to compensate for tilt so that sudden shakes do not drastically disturb the point cloud. Calibrating the IMU and verifying that the electronic compass is correctly oriented before the job increases trust in the sensor outputs. By doing so, you can secure a certain level of positional accuracy during capture, requiring only minor adjustments in post-processing even without markers.

By fully leveraging positioning and attitude sensors such as GNSS and IMU, you can digitally replace the role that physical markers traditionally played in alignment. As a result, the likelihood of meeting required accuracy without placing markers on site increases significantly.

Tip 5: Careful post-processing and accuracy verification

How you handle data in the office during post-processing determines the final quality of deliverables. When integrating point clouds and photos captured without markers, perform as much optimization as the software allows. In photogrammetry software, run bundle adjustment (Bundle Adjustment) on the captured images to distribute and minimize errors across the dataset. During preview, check for areas where automatic matching has misaligned images and, if necessary, add manual tie points rather than relying solely on automatic processing. In laser scan registration, review the registration error between scans reported by the software and confirm it falls within acceptable limits. If large misalignments are detected, set additional control points or exclude problematic scans. Use global adjustment functions where available to minimize errors across all scans and verify that local misalignments have not compounded into global distortion.

After data integration, clean up noise and unnecessary points. Remove spurious point cloud noise and duplicate points, and filter or smooth sections of the terrain that should be flat if stepped artifacts appear. Be careful not to over-smooth edges or fine structural details. When combining datasets captured on different days, you may see subtle differences in color or point density; if appearance is an issue, perform texture color correction and resample to even out point density.

Always perform an accuracy check on the final 3D model or point cloud. One method is to compare distances between two points in the model with field-measured values. Compare dimensions known from tape measures or existing drawings (for example, column spacing or road width) to corresponding distances measured on the point cloud to confirm differences are within tolerance. Comparing multiple locations helps determine the model’s overall scale accuracy. If possible, measure a few points independently with a surveying instrument and perform a “checkpoint validation” by comparing their coordinates with the model’s corresponding points to objectively evaluate the absolute accuracy (the offset relative to the reference coordinate system) of the markerless model. If these checks reveal insufficient accuracy, apply scale corrections or translate the model based on known points as needed. Some photogrammetry software allows you to input known distances or coordinates after initial processing to re-optimize the model—use these features to reduce errors.

Accuracy verification completes the surveying workflow. Making careful post-processing and validation routine ensures that markerless measurements can be relied upon. Feedback from verification—insights on where errors occurred in the process—should be used to refine subsequent measurement plans and continuously improve accuracy.

Conclusion

Markerless 3D measurement is becoming practically viable thanks to a variety of techniques and the application of the latest technologies described above. With proper planning and preparation and by acquiring data using appropriate methods, cases where control point placement was once deemed essential can now often be completed with sufficient accuracy. Omitting marker placement shortens work time and removes the need to install markers in dangerous or hard-to-reach locations, improving safety.

As technology advances, the range of applications where markerless high-accuracy measurement is possible will continue to expand. One representative example is the high-precision GNSS device "LRTK" that can be attached to an iPhone. By attaching LRTK to a smartphone, centimeter-level positioning (half-inch level) can be achieved with a handheld device, allowing photogrammetry and LiDAR scan data to be tagged with high-precision positions. Using such modern tools makes it possible to meet accuracy requirements that previously required dedicated surveying instruments without placing markers on site.

Markerless 3D measurement is rapidly becoming a new standard in surveying and inspection practice for construction and civil engineering. Please refer to the tips and tools introduced here and take on markerless 3D measurement in a safe and efficient way.