What is deformation inspection using point clouds? Four perspectives for checking cracks and settlement

In condition surveys, how accurately site conditions are recorded and how objectively they can be compared greatly affects the reliability of subsequent repair decisions and maintenance management plans. Conventional condition surveys have relied mainly on visual inspection, photographs, sketches, and checks using tape measures or staffs, but when the object’s shape is complex, when a large area must be captured in a short time, or when reliable comparisons over time are required, the reproducibility and quantitativeness of the records tend to be limited.

What is attracting attention is deformation surveys that utilize point clouds. A point cloud is data that records the surface shape of an object in three dimensions as a collection of many points, and its strength lies in the ability to preserve structures, terrain, slopes, pavement surfaces, retaining walls, tunnel interiors, and building exterior walls not as surfaces but as high-density spatial information. This makes it easier to later interpret sensations of irregularity that were perceived on site as dimensional or shape differences.

Many practitioners who search for "point clouds for deformation surveys" are not simply looking to learn a new measurement method; they want to know whether point clouds can actually be used to verify deformations, how far they can detect important items like cracks and settlement, and what to focus on during implementation to reduce the risk of failure. Especially in maintenance and inspection work, because survey results feed into reports, repair design, procurement decisions, and long-term monitoring, merely producing attractive three-dimensional data is not sufficient. What matters is preparing the data so that the presence and degree of deformation can be read.

This article organizes the basics of condition assessment using point clouds, then explains how to view representative distresses—cracks, settlement, deformation, delamination, and missing parts—from four perspectives. It also delves into approaches for achieving results in actual practice and into common on-site failures and countermeasures. Please read this as a concise, practical overview that enables practitioners who want to leverage point clouds for condition assessment to understand the practical concepts they should grasp before implementation.

Table of Contents

• What is a deformation survey using point clouds?

• Points to check when inspecting cracks

• Points to check during settlement verification

• Key points to check when inspecting deformation and deflection

• Points to check when inspecting for peeling, defects, and surface changes

• Procedures for conducting deformation surveys using point clouds

• Common Pitfalls and Countermeasures When Using Point Clouds

• Summary

What is a deformation survey using point clouds?

A deformation survey using point clouds is an inspection method that densely acquires the three-dimensional shape of a target and, by interpreting shape differences, positional offsets, surface irregularities, local defects, and gradient changes, identifies abnormalities in structures and ground. It is not merely a three-dimensional record; the essence is to prepare the data to a state where it can be used to detect, understand, compare, and explain deformations.

One major advantage of point clouds is that they can capture and preserve the site’s geometry broadly, in fine detail, and as surfaces. For example, visual inspection or photographs can produce biased records depending on the area and angle photographed. With contact measurements such as a tape measure, you may be able to measure required points in detail, but information about parts you later want to review from a different perspective may not be retained. In contrast, point clouds tend to preserve surrounding geometry, including areas not anticipated at the time of the survey, and in post-processing you can cut cross-sections, examine differences, or extract only specific ranges. This ability to “re-check later” is a characteristic that makes them highly suitable for condition surveys.

On the other hand, having a point cloud does not mean you can know everything. Point clouds are strong for information about surface geometry, but they do not directly indicate internal conditions such as internal cavities or rebar corrosion. Likewise, features with very small widths, such as cracks, are affected by measurement density, the condition of the target surface, lighting conditions, observation angle, and noise level. In other words, point clouds are not a universal substitute; you must clearly define what you want to observe and design in advance the accuracy, density, reference coordinates, and acquisition range that meet that purpose.

There are broadly two situations in condition surveys where point clouds come into play. One is when you need to capture the shape over a wide area to make it easier to detect signs of abnormalities. Representative examples that are difficult to grasp with only local dimensional measurements include overall bulging of a slope face, local deformation of a retaining wall face, settlement of deck or pavement surfaces, and changes in the cross-sectional shape of a tunnel interior. The other is when the same object is measured at different times and compared as differences. If the point clouds from the previous and current surveys can be compared in the same coordinate system, it becomes easier to determine the amount of displacement and the direction in which the deformation is progressing. In maintenance management, what matters is not only what the condition is now, but also how much it has progressed.

In practice, it is important not just to view point clouds as they are, but to translate them into readable forms such as cross-sections, elevation distributions, surface slope, distance differences, and time-series differences. For example, subsidence becomes easier to understand as a height difference, deformation as a bulge or depression relative to a reference surface, and loss as a break in surface continuity or the amount of missing material. By organizing these interpretation methods, survey results become less dependent on the individual analyst’s experience and turn into deliverables that are easier to explain.

Moreover, condition surveys using point clouds sit somewhere between inspection and surveying. This is because they can handle both the visual understanding required for inspections and the quantitative positional and shape information required for surveying. For that reason, they are advantageous in that they make it easy to share information among multiple stakeholders—such as maintenance departments, surveying firms, construction managers, and municipal officials. Even in situations where judgments tend to differ when relying on photographs alone, if cross-sections or differences based on point clouds can be shown, it becomes easier to establish a common basis for discussion.

However, to realize this advantage in practice, you must design the entire workflow—from on-site data acquisition to coordinate management, analysis, and the standardization of comparison conditions. What a deformation survey requires is not merely taking measurements, but establishing comparisons that are reproducible. In the following chapters, we will sequentially delve into the four perspectives that are particularly important in deformation surveys.

Key points to check when inspecting cracks

One of the topics that attracts the most interest in condition assessment using point clouds is crack inspection. What practitioners worry about first is whether cracks can really be seen in point clouds and to what extent they can be confirmed. In short, point clouds are effective for identifying the presence of cracks and mapping their locations, but they cannot reliably interpret every crack width with high precision. Misunderstanding this leads to a mismatch of expectations after implementation.

The first thing to check when inspecting cracks is whether the linear shape of the crack itself appears in the point cloud. If the target surface has groove-like changes or steps, and the measurement density is sufficient, traces of those can appear in the point cloud. In particular, on concrete, mortar, or stone surfaces, when a crack manifests as an irregularity in surface geometry, it can be easier to detect depending on observation conditions. However, very fine hairline cracks with small width and shallow depth can be difficult to detect reliably using only a point cloud. Therefore, it is more realistic to consider point clouds not as the primary tool for crack inspection but as being strong for recording positions, understanding distributions, and organizing relationships with surrounding geometry.

Next, it’s important not to look at a crack in isolation but to view it together with the displacement and deflection of the surrounding surfaces. A crack is not merely a surface blemish; it can be related to underlying differential settlement, bending, extrusion, thermal stress, drying shrinkage, changes in support conditions, and so on. The advantage of point clouds is that they can simultaneously record the surface geometry in front of, behind, and to the sides of the crack location. For example, if there is an oblique crack on a wall surface, you can broadly check whether there is slight bulging of the surface or misalignment at member ends in the surrounding area. Because photographs alone may reveal the presence of a line but often fail to capture the three-dimensional condition of the surrounding surface, this is a strength unique to point clouds.

In crack inspection, the thinking about measurement direction and illumination conditions is also important. Measurements taken only normal to the surface can make changes in grooves and steps appear less pronounced. Conversely, acquiring data from multiple directions while varying the angle can make shading differences of the shape more likely to appear in the data. In other words, if you use point clouds for crack inspection, you need to plan in advance how to observe the surfaces where cracks are likely to occur, not just collect data broadly. In particular, surfaces such as walls, the underside of deck slabs, tunnel linings, and retaining walls can yield greatly different results depending on how you approach the target surface.

Furthermore, the concept of time-series comparison is highly effective for crack inspections. Subtle changes that are difficult to interpret from a point cloud captured at a single time can become apparent when comparing the previous and current data, revealing progression in the opening of surfaces around the crack or in step offsets. In practice, it is necessary not only to measure the crack width at a single point but also to comprehensively consider the crack’s extension direction, displacement of the surrounding surface, and the positional relationship between the upper and lower parts. Point clouds serve as the foundation to support this comprehensive assessment.

One thing to be careful about is not to mistake noise for cracks. On rough surfaces, dirt, deposits, wetness, vegetation, highly reflective surfaces, or under strong shadow conditions, the point cloud can show disturbances. If these are taken as cracks as they are, it can lead to misreading. Therefore, when confirming cracks, you need to check not only local linear disturbances but also continuity with the surroundings, reproducibility from different angles, consistency with photographic records, and the presence or absence of offsets when viewed in cross section. Because point clouds give a strong impression of quantitative data, they are prone to being overtrusted, so a skeptical view of data quality is indispensable.

In reporting practice, it is also important to organize and indicate where cracks are located across the entire surface. What makes point clouds effective is that they make it easy to get an overview of where abnormalities are concentrated within the whole subject. If you organize the crack distribution with consideration for positional relationships where stress tends to concentrate—such as structural edges, around openings, near joints, and around supports—you produce material that leads to cause estimation rather than a mere record of discoveries. In other words, when confirming cracks with point clouds, the essence is to look at the surface as a whole, not just the lines.

Points to Consider When Checking Settlement

Settlement detection is a domain where the benefits of point cloud use tend to be relatively easy to see. Compared with fine linear changes such as cracks, settlement more readily appears as a change in elevation, making it easier to use the point cloud’s three-dimensional coordinate information directly. In locations where settlement effects are a problem—floor surfaces, pavement surfaces, around foundations, ground surfaces, around gutters, around manholes, the tops of retaining walls, sedimentation basins, and around work platforms—area-based assessment using point clouds is highly effective.

The first thing to keep in mind when checking for settlement is not to judge based solely on simple elevation differences. Sites have original slopes and design-intended inclines. For example, if a paved surface is slightly pitched for drainage, you must not mistake that pitch for settlement. What matters is to assess, taking into account the design-assumed gradients and continuity with the surroundings, whether there are any unnatural localized depressions or undulations. Because point clouds capture a wide area at once, they make it easier to judge anomalies not only from local sinking but also in the context of their connection to surrounding areas.

Another important point is to regard settlement as a surface phenomenon rather than as a point. Conventional leveling and surveys of individual points are excellent for accurately determining the elevation of selected points, but they can make it difficult to detect shape changes between points. With point clouds, for example, it becomes easier to confirm on a surface whether only the center of the pavement has settled slightly, whether there is rut-like settlement along the wheel paths, or whether only the area next to a structure has sagged. In condition surveys, the shape of this settlement is important, because knowing the shape makes it easier to infer whether the cause is a localized loss of bearing capacity, poor drainage, or the influence of surrounding loads.

In subsidence assessment, how you establish the reference has a major impact on success or failure. Point clouds are useful if you only want to see relative height differences, but for comparisons over time and for consistency with other sources you need to tie them to stable reference coordinates. If the coordinate systems differ between the previous survey and the current survey, apparent subsidence or heaving may occur. Therefore, when discussing subsidence you must confirm whether the datasets have been aligned to the same reference, whether a sufficient number of fixed reference points for comparison have been secured, and whether the overall surrounding area is consistent. In practice, management of these references is surprisingly easy to overlook, and even if point clouds are acquired, the data can become unusable for comparison.

Also, settlement does not necessarily occur in isolation and can be linked with cracking and deformation. For example, if one side of a structure settles, tensile cracks may develop in the upper portion or out-of-plane tilting may occur. When confirming settlement with point clouds, instead of simply looking at height alone, combining that with the arrangement of surrounding members and surface distress enables more practical judgment. In particular, for ongoing monitoring, attention should be paid not only to the absolute amount of settlement but also to the expansion of the settlement area and changes in the displacement gradient. Whether the affected area is widening or the step is becoming steeper will change the priority of countermeasures.

To make point clouds useful for settlement verification, it is also important which representation is used during the analysis stage. Viewing by cross-section, by elevation differences, by distance from a reference plane, or by time-series differencing—each representation affects how easily site stakeholders can understand the results. When using point clouds in reports, you should not simply present a three-dimensional view; you need to organize the output so that it is clear where the settlement locations lie in plan and in which direction and by how much they have changed. Because point clouds contain a large amount of information but can easily become data that fails to communicate if shown incorrectly, adopting perspectives that reinterpret the data according to the investigation objectives is essential.

Points to check for deformation and deflection

The third important perspective in condition assessment is checking for deformation and deflection. Cracks and settlement are relatively easy to visualize, while phenomena such as overall structural sagging, out-of-plane deformation, bulging, and twisting can be difficult to judge from photos or visual inspection alone. Point clouds are particularly effective for confirming deformations because they make it easy to capture such shape changes that span wide areas both as surfaces and in cross section.

For example, for a retaining wall, it is necessary to look not only for the presence or absence of localized cracks but also whether the entire wall face has moved forward, whether the top edge line is wavy, or whether the inclination changes between the upper and lower parts. For a tunnel, there are situations where you want to check for flattening or asymmetry of the internal cross-section, and for local deformations of the sidewalls or the lining surface. For floor slabs, beams, and walkway members, the amount of deflection at the center and any left–right difference are important. Because such deformations cannot be understood by measuring only individual points and are difficult to quantify by visual inspection alone, an approach that captures the overall shape with a point cloud and then examines differences from cross-sections or reference planes is effective.

What's important in deformation verification is to clarify what you use as the reference for considering something a "deformation". For structures that originally have curved surfaces or slopes, evaluating solely by the difference from a simple plane can be misleading. Only by appropriately setting the comparison reference—such as the design geometry, the baseline shape at the time of installation, surrounding surfaces that can be regarded as intact, or point clouds from past time points—can deformation amounts be treated as meaningful numbers. Especially for older structures, where records of the as-built shape are often insufficient, it is important to determine on site which areas to standardize as stable references.

Also, when checking deformation, it is necessary to consider local deformation and overall deformation separately. Local deformation refers to small changes such as only a specific part bulging, denting, or a corner chipping. Overall deformation refers to changes such as an entire member sagging, an entire wall tilting, or an entire cross-section contracting. The strength of point clouds is that both can be seen in a single dataset. During inspections, local abnormalities tend to catch the eye, but what is truly dangerous is often the progression of overall deformation. For example, a small chip in part of a wall may itself be a superficial problem. However, if the entire face of the same wall is being pushed forward, one should suspect back pressure or a change in support conditions. Point clouds help clarify the relationship between these local and overall deformations.

Time-series comparisons are also very effective for confirming deformation. A single measurement only reveals the shape at that point in time. However, the significance of deformation changes greatly depending on whether the phenomenon is progressive. Even if the displacement from the past is small, if it is continuously increasing, caution is necessary. Conversely, if a certain amount of deformation has occurred but has been stable for a long period, the priority of response may change. By continuously acquiring point clouds in the same coordinate system and performing cross-section and surface-difference comparisons, it becomes easier to understand where and in which direction movement is occurring. This good compatibility with continuous monitoring is a major value of deformation surveys using point clouds.

Even in deformation inspection, incorrect data acquisition leads to wrong judgments. If the acquisition range is too narrow, there will be insufficient undeformed reference areas, making it difficult to read relative deformation amounts. Also, if the viewpoint is biased, a protrusion visible from one direction may not be adequately reproduced from another. Therefore, when the purpose is deformation confirmation, it is important to capture a wider area that includes not only the target itself but also the surrounding stable and continuous parts. Deformations are hard to understand if you only crop out the abnormal portion; their meaning only becomes clear in relation to the surrounding context.

Points to check when inspecting for delamination, missing parts, and surface changes

The fourth perspective is how to capture changes in surface condition such as delamination, chipping, wear, surface roughness, and changes in surface geometry. These are not always described with as clear keywords as cracks or settlement, but they are very important in maintenance and management practice. In particular, whether surface delamination of concrete structures, chipping of stone, slope erosion, pavement surface wear, or surface deterioration of retaining walls and exterior walls can be detected early will affect the accuracy of repair planning.

When examining this type of deterioration with a point cloud, the basic principle is to focus on surface continuity. Intact surfaces, while exhibiting roughness depending on the material and construction condition, are generally perceived as continuous shapes. In contrast, when there is delamination or material loss, the surface may be locally interrupted, dented, or have its contour disrupted. Because point clouds can capture such surface losses as spatial forms, they make it easier to grasp the amount of indentation and the extent of missing material that are difficult to discern in photographs. In particular, corner and edge losses, damage around member joints, and depressions left after detached fragments have fallen are easier to explain when organized three-dimensionally.

Also, when checking surface changes, understanding the areal extent is important. Delamination and wear can occur at a single point, but more often they occur preferentially in response to specific flow or loading conditions, rain exposure, freeze–thaw effects, frequency of contact, and so on. With point clouds, for example, it is easier to get an overview of distributions such as whether deterioration is concentrated along the lower edge, whether localized defects are scattered in the central area, or whether roughening is progressing along member boundaries. In condition surveys, these distributional characteristics help with cause estimation and planning the next monitoring program.

When checking for surface changes, be careful to distinguish roughness inherent to the material from roughness caused by deterioration. Stone, sprayed surfaces, areas with exposed aggregate, and finishes that are inherently uneven can make the surface shape appear irregular. Treating all of that irregularity as anomalous leads to false positives. Therefore, those who read point cloud data need a prior understanding of what surface characteristics the object normally has. While comparing with the surface pattern of intact areas, it is important to check whether there are locally different types of roughness or any loss.

Furthermore, evaluations of delamination and material loss must also take safety implications into account. While cracks and settlement relate to assessments of progression, delamination and material loss can directly lead to falling debris or contact accidents. Therefore, changes identified in point clouds should be linked not only to discussions of dimensions and volume, but also to management information such as where they are located, how far they are from passageways or areas in use, and whether there is spread to similar surrounding locations. Point clouds contain rich shape information, but to make that useful for safety management decisions, you need to organize it as positional information.

When checking surface changes, comparing different times is particularly effective. At a single point in time you may not be able to tell whether a chip has been there for a long time or a delamination has recently expanded, but comparing with past point clouds will reveal increases or decreases. In maintenance practice, whether a condition is progressing is more important than the mere presence of a defect. Because point clouds can be compared within the same spatial frame for past and present, it is easier to maintain a continuous basis for decision-making even when personnel change.

Procedure for Conducting Deformation Surveys Using Point Clouds

To successfully carry out condition surveys using point clouds in practice, it is important to design a workflow that aligns with the survey objectives before selecting measurement equipment or analysis methods. Here, we outline the basic workflow that practitioners should be familiar with.

The first thing to do is to clarify what you want to check in the survey. Whether the objective is to understand the distribution of crack locations, to determine the amount of settlement, to monitor long-term deformation, or to preserve the overall shape for reporting will change the required density, accuracy, and acquisition range. A common mistake in condition surveys is to "collect point clouds for the time being" while the purpose remains vague. That often results in insufficient resolution later or missing reference points needed for comparisons.

Next, decide which part of the object you will treat as the reference. To discuss deformations, you need parts that can be regarded as stationary or a reference that can be shared with the previous survey. For example, if you capture part of the structure or stable surrounding areas, it will be easier to align positions in post-processing. Conversely, if you narrowly capture only the area showing abnormalities, comparison and evaluation become difficult. It is important not to forget the premise that deformations should be interpreted in relation to their surroundings.

In on-site acquisition, it is important to be mindful of reducing blind spots and of capturing the target surface from multiple directions. The appropriate way of capturing differs somewhat between local defects such as cracks and spalls, and global deformations such as settlement and sagging. If you prioritize local confirmation, a close sampling density is required, whereas if you are looking at overall deformation, the relationship with a wide area is important. In practice both are often required, so it is effective to combine approaches for wide-area overview and detailed inspection.

After acquisition, we carefully perform preprocessing such as removal of unnecessary points, target extraction, noise filtering, alignment, and normalization. If this processing is done sloppily, the difference and cross-sectional results will become unstable. In deformation surveys, it is more important to prepare the data in a comparable state than to make it look clean. In particular, for time-series comparisons, it is necessary to record the range over which alignment was performed and which part was used as the fixed reference.

During the analysis phase, choose the perspective according to the subject. For cracks, consider the relationship with surrounding surfaces; for settlement, look at elevation differences and changes in slope; for deformation, examine cross-sectional differences and deviations from reference planes; for delamination or missing parts, focus on local depressions and changes in surface continuity. The important point here is not to present the three-dimensional display as-is, but to convert it into a form that stakeholders can easily assess. Only when field staff, managers, and the client can view it from the same perspective, organized in this way, will point clouds function as practical documentation.

Finally, carry out back-and-forth checks with on-site verification. Do not try to rely solely on the point cloud; by cross-referencing site photos, close-up visual inspections, existing drawings, and past inspection records as needed, the reliability of your assessments increases. Point clouds are a powerful recording tool, but they are not a substitute for on-site judgment; they are the foundation that supports decision-making. Maintaining this correct perspective prevents failures after implementation.

Common Mistakes and Countermeasures When Using Point Clouds

Point cloud-based condition surveys are effective, but especially in the early stages of adoption, failures in which "we have the data but it doesn't lead to decisions" are likely to occur. Here we summarize the failure patterns that are particularly common in practice and their countermeasures.

The first mistake is misjudging the required accuracy and density. A common mismatch is having only a coarse point cloud when you need to see cracks, or conversely collecting high-density data only locally when you need to observe widespread subsidence, leaving the overall picture unclear. As a countermeasure, anticipate in advance how each type of deterioration needs to be visualized, and treat local checks and overall checks separately. If the accuracy requirements for the objective are not clarified, rework after acquisition is likely to occur.

The second issue is insufficient control of reference points for comparison. If the coordinate conditions between the previous and current surveys are not consistent, comparisons of settlement and deformation cannot be made. If you are aiming for time-series monitoring, you need to ensure that measurements can be taken using the same reference each time. In condition surveys, this reference control is one of the most important parts, despite being unglamorous.

The third is confusing noise with actual defects. Water, reflections, shadows, vegetation, deposits, and the presence of people or vehicles can all introduce disturbances in point clouds. Treating local disturbances as anomalies will lead to false detections. Countermeasures include checking from multiple directions, verifying continuity with surrounding areas, and thoroughly cross-referencing with site photographs. When interpreting point clouds, you need to be as skeptical about data quality as you are about searching for anomalies.

The fourth issue is that people are satisfied with acquiring 3D data but fail to translate it into a form usable for reporting and decision-making. Point clouds contain a lot of information, but as they are they can be hard for stakeholders to understand. In practice, unless you organize them into cross-sections, differences, planimetric positions, area of interest, and displacement directions, the valuable data will not be put to use. The countermeasure is to work backwards from the final deliverable and decide up front how the data needs to be presented.

The fifth is trying to make all judgments based solely on point clouds. There are things point clouds alone cannot reveal sufficiently, such as internal deterioration, material degradation, or fine surface conditions. Point clouds excel at broad-area understanding and shape comparison, but there are also cases where on-site close inspection or combining with other methods is necessary. Avoid treating point clouds as omnipotent; using them in the right place for the right purpose will ultimately lead to the greatest effectiveness.

Summary

Point-cloud-based condition surveys are a method that records site conditions three-dimensionally and as surfaces, making it easier to understand defects such as cracks, settlement, deformation, delamination, and loss in terms of both their location and shape. In particular, the ability to capture large-area geometry at once, to review it later from different viewpoints, and to easily perform time-series comparisons in the same coordinate frame are major advantages not provided by traditional visual- or photo-based surveys.

On the other hand, point clouds are not a magic tool that will automatically reveal deterioration simply by being introduced. You must clearly define what you want to check and design the density, accuracy, acquisition range, and reference coordinates to suit that purpose; otherwise they will not provide data usable for decision-making. If you are inspecting cracks, consider their relationship with surrounding surfaces; if you are inspecting settlement, look not only at absolute elevation values but also at surface connectivity and changes in gradient; if you are inspecting deformation, check both local and global conditions; if you are inspecting delamination or loss, pay attention to surface continuity and changes over time.

To make deformation surveys truly useful in practice, the notion of preserving measurements as comparable data is indispensable. Being able to align the previous and current surveys under the same criteria, to explain where abnormalities are occurring on site, and to continue monitoring even when personnel change — these factors determine the quality of maintenance management.

In that sense, using point clouds is not merely a matter of three-dimensionalization but an initiative that changes how field information is handled. Especially in situations where you want to streamline on-site position checks, identification of reference points, revisits to deteriorated locations, and sharing of recorded positions, creating an environment where coordinates can be quickly verified is important. In these day-to-day deterioration inspections and maintenance sites, leveraging LRTK, an iPhone-mounted GNSS high-precision positioning device, makes it easier to conduct on-site coordinate confirmation and stakeout more nimbly. By combining shape capture through point clouds with the use of high-precision positional information, the recordability and reproducibility of deterioration inspections can be further enhanced. Looking ahead to continuous inspections and labor savings, organizing not only point clouds but also the means to reliably secure positions on site will be the practical differentiator.