Application of TLS Technology for Documentation of Brickwork Heritage Buildings and Structures

Abstract

Remote measurement of historic buildings and structures using the technology of terrestrial laser scanning (TLS) is becoming a more and more popular approach for conducting inventory activities, documentation and conservation works. In this paper, TLS was harnessed for analysis of historic brickwork structures from the 19th century. During the research programme, chosen brickwork heritage buildings were scanned. Based on the collected geometric data of the point cloud, it was possible to create an exact model of the scanned objects. The obtained radiometric information of the point cloud allowed us to identify changes in the surface of walls, such as cavities, cracks and previous repairs. Moisture was also identified in some cases. The conducted tests enabled the identification of brickwork in need of urgent repair. It was possible to assess the general technical state of the tested structures. The possibilities and limitations of the TLS diagnostic methodology of analysing the quality of historic brickwork and its future possible applications were indicated.

1. Introduction

Laser scanning, which represents one family of light detection and ranging (LiDAR) technology, has been known and used mainly for geodesic applications since the 1980s [1,2]. Geodetic measurements conducted with the help of terrestrial laser scanning (TLS) proved to be very useful and enabled fast development of the technology. TLS is an apparatus that utilizes a laser beam for remote measurements. It acquires datasets through numerous measurements. Each single scanned point by a TLS apparatus is described by a set of data which includes distance (r), horizontal angle (ϑ), vertical angle (φ) and so-called intensity of the captured signal. The datasets in question are used for creation of both 2D and 3D digital models. Over time, new types of TLS (based on different wavelengths of the laser beam) and different scanning methodologies (time of flight, phase-shift) have become available. Currently, TLS is a quite popular and reasonably affordable tool used globally in civil engineering. Apart from purely geodetic applications, TLS is increasingly used for other civil engineering applications. The non-geodesic applications cover a wide range of construction activities associated with façade analysis [3,4], monitoring of bridges [5,6], monitoring of structure deformations [7,8] (including dams [9,10] and tunnels [11,12]) and following landslides [13,14]. It was proven that in all the above applications, TLS is a very efficient, flexible and reliable tool.

Commercially available geodetic TLS can also be used for more sophisticated purposes such as assessment of saturation of building materials [15,16], moisture movement in walls [17] and assessing roughness of pavement [18]. According to some researchers, utilization of LiDAR techniques during the colonization and “commercialization” of the Moon and Mars is inevitable [19]. Taking into account all the above facts, the authors decided to use TLS for assessment of the structural health of historic buildings.

Terrestrial laser scanning technology for documentation of brickwork heritage buildings has been tested by scientists for many years. The conducted research programs covered such areas of possible applications as:

• Application of TLS in moisture detection in heritage buildings [20].
• Combined radiometric and geometric analysis of TLS data for heritage site preservation [21].
• Surface fractures and materials behaviour of cultural heritage buildings based on the point cloud [22].
• Integration of photogrammetric and terrestrial laser scanning techniques for heritage documentation [23,24,25].
• Three-dimensional digital documentation of cultural heritage site based on terrestrial laser scanning and unmanned aerial vehicle photogrammetry [26].
• Harnessing of terrestrial laser scanning in the maintenance of historic buildings [27].

Keeping in mind the above achievements, the authors decided to conduct a research programme dedicated to a specific type of heritage buildings. This paper presents exemplary TLS tests of 19th century brickwork heritage buildings and structures. Such architectural objects are very popular in Europe and are notorious for being difficult and expensive to efficiently renovate. Using TLS enables acquiring full knowledge about the quality of the brickwork of a scanned building. During the research programme, it was demonstrated that identification of missing bricks, missing mortar and detection of moisture in a scanned brickwork wall are reasonably easy to conduct. The proposed approach enables detection of key areas of brickwork which are in urgent need of repairs. All the assessment is performed quickly, remotely and non-destructively, which is very important in the case of 19th century heritage buildings (e.g., churches or military structures) [28,29]. These research programmes were based only on using geometric (x,y,z) data provided by a scanner. The main aim of the current work is to indicate the possibility of combining geometric (x,y,z) and radiometric (intensity) data from TLS measurements in the context of inventory and assessment of historic brickwork heritage buildings and structures. The conducted research shows the general outline of the proposed concept, supported by examples of point cloud analyses based on spatial coordinates (x,y,z) and intensity values. The innovative character of the conducted research programme is based on harnessing information sourced from different sensors. The authors simultaneously used geometric information (xyz), radiometric information (intensity) and digital image information (image RGB) to assess the technical state of the historic brickwork structure. The proposed approach can be further developed by adding information from thermal images. This paper has also a review character and refers to numerous references to keep the best possible context of the discussed methodology to the state of the art.

2. Materials and Methods

The methodology, which was developed and tested during previous research programmes [30,31] and harnessed by the research team, is comprised of four main stages. The sequence of four main stages (1—planning the measurement; 2—TLS data acquisition; 3—pre-processing of data; 4—post-processing of point clouds) is presented in Figure 1. Each of the stages covers certain activities which finally lead to building 3D models of a scanned heritage brickwork structure. Stage 1 covers planning of measurement and generally depends on the tested object. The minimum number of scanning positions providing full coverage of the scanned object should be found with a certain level of data quality. Scanning time is very often known as a critical factor of data acquisition. Increasing the scanning positions can provide more detailed data, but it can also cause data redundancy of dataset. Trade-offs must be found between the scanning time and data qualities. An overview on the planning of measuring TLS stations can be found in [32,33]. Stage 2 covers TLS data acquisition. The spatial resolution is determined as a function of the distance between the TLS and the target and the set vertical and horizontal angular resolution. The distance and resolution should fit in such a way as to obtain similar values of the spatial resolution of the scanned object from different scanning stations. An increase in the scanning resolution causes a significant increase in scanning time. Therefore, the scanning parameters should be wisely selected in the context of the expected detail of data acquisition. Using phase-shift TLS (in opposition to the time-of-flight TLS), it is possible to obtain data at a high speed of over 2 million points/s, which allows for a relatively fast measurement. Apart from the RGB camera, the scanners can have additional sensors, such as an IR camera, which can also be used for documentation [34]. While surveying, it is worth collecting data at the same time using several sensors (point clod, RGB image and thermal image). During stage 3, pre-processing of data regarding the registration of point clouds to one coordinate system and data filtration of noise from point clouds are executed. Data registration is an automated process that uses various approaches such as registration based on special artificial targets, point-to-point, cloud-to-cloud, plane-to-plane or mixed method [35,36]. Scanner manufacturers usually provide various data-filtering methods in their software, and this process is fully automated as well. The most popular data filtration methods are well-described in [37,38]. Finally, during stage 4, the post-processing of point clouds takes place. It is the most complex and time-consuming process of the whole testing procedure. The main aim is to build a three-dimensional model of the object based on 3D coordinates using CAD software. Defect detection on an object can be performed based on spatial coordinates and intensity value of point clouds. Additionally, information in the RGB image and thermal image may be very useful in this step. Surface changes such as moisture, weathering, peeling and bio deterioration can be identified using intensity of point clouds, RGB image and thermal image information. This research can also be significantly automated using deep learning [39,40,41,42] and other various automatic and intelligent methods for defect detection [43,44] through point clouds and image analysis. Other details regarding methodology and metrics of the presented research programme are presented in Appendix A.

Use of the 3D model deviation analysis of building walls was performed, enabling detection of cracks and defects (including bio-deterioration, moisture penetration, peeling, delamination, discoloration and changes in roughness of bricks surface).

The Z+F IMAGER 5016 TLS was used for conducting the scanning. This apparatus and its non-geodetic applications were thoroughly described by the research team in a previous publication [45].

3. Results and Discussion

During the research programme, two different brickwork heritage buildings were scanned. The first one was a brick citadel around the Kosciuszko Mound in Cracow, Poland. The citadel around the Mound was built between 1850 and 1854. Currently, some parts of the citadel are in a poor technical state (Figure 2). The scanning was conducted from a distance of 17 m using a super high resolution of the scanner.

Using data acquired by the TLS, a computation of the distance between a point cloud of the building’s wall and a reference plane was conducted. This is a common approach to deviation analysis [46]. The reference plane is determined from the points belonging to the wall using the mean sum error (MSE) method presented by Chen [47]. The deformation maps of the wall are then created. In Figure 3, the deviation analysis of the citadel wall is presented.

In Figure 4, building of the 3D model is summarized. The model was built based on the Poisson Surface Reconstruction (PSR) method in CloudCompare software. The triangular mesh generation algorithm proposed by Misha Kazhdan of Johns Hopkins University was used in PSR [48]. One can observe that the achieved precision of scans enables identification of missing pieces of bricks, mortar and general quality of the scanned wall. The defect detection procedure (based on assessment of distance of points from the vertical plane) is demonstrated in Figure 5. Firstly, two types of used materials are clearly differentiated (concrete foundation wall and brickwork wall). Secondly, three missing bricks are identified. Both identifications are executed using the deviation analysis. The detected cavities are analysed further. Their dimensioning is conducted. An exemplary procedure of dimensioning of cavity no. 2 is presented in Figure 6.

Defects, cracks or surface changes are dimensioned based on 3D coordinates of the point cloud. The accuracy of dimensioning or creating 3D models depends mainly on the error of determining the coordinates (x,y,z) of points and the density of the point cloud. Point measurement error depends on TLS technical specifications, such as precision of distance measurement, vertical angle accuracy and horizontal angle accuracy, laser spot size and laser beam divergence (see Appendix B). Other factors such as weather conditions of the measurement, the type of surface scanning (colour, roughness, saturation) or the accuracy of point clouds registration in the one coordinate system also affect the accuracy of the final 3D models. In general, it can be assumed that the accuracy of dimensioning or building 3D models is a few millimetres. You can find many studies that present analysis of the accuracy of terrestrial laser scanning measurements [49,50,51].

Detection of other types of points of interest in the wall are also possible. In Figure 7, an exemplary detection of biological corrosion of the wall is presented. In this case, the analysis of the intensity of the received laser beam is crucial. Intensity is defined as the relationship between the emitted and received signals’ power during laser scanner measurement. It should be noted that each surface exhibits a different capacity to reflect an impulse, which depends primarily on its physical and chemical properties. Changes in colour, roughness and surface humidity significantly affect the absorption and dispersion of the laser beam. Thus, this property can be used to identify bio-deterioration, weathering and peeling of the building wall. However, changes in two parameters at the same time, e.g., roughness and humidity, make it difficult or impossible to detect these changes.

The value of intensity (scalar field) of a dry, stone surface is equal to 0.345. The surface of the same stone covered by moss is equal to 0.183. The border between both areas of the stone surface is clearly visible on the image showing both a traditional photograph and the image showing values of intensity.

A similar procedure can be used for identification of different types of bricks or places which are prone to saturation. In Figure 8, an exemplary area of the wall is presented where both types of the phenomenon are present. Layers of two different types of bricks can be easily identified. In Figure 8, a dashed line marks the place of changing the type of bricks used during the reconstruction. Each change on the surface (physico-chemical properties) affects the absorption and dispersion of the laser beam which translates into the obtained value of intensity. Under the windows, stains created by rainwater (probably limestone raids that have settled on the bricks) are clearly visible. In Figure 8, randomly chosen bricks representing both types of bricks were chosen (Brick 1 and Brick 2). The detailed analysis of values of intensity for both bricks is presented in Figure 9.

Two types of bricks are characterized by a different colour and roughness, which translates into different absorptions and dispersions of the laser beam. These differences are mirrored by the distribution of values of intensity shown in the histogram.

The difference in intensity of the two bricks was significant and equal to 0.45. The second analysed brickwork heritage building was located in Olsztyn (Kortowo) in Poland. The building was originally erected in 1886 as a part of the Provincial Mental Sanatorium Kortau. In Figure 10, a deviation analysis of a wall with multiple defects is presented.

Again, the deviation analysis enables the building of a precise 3D model. The 3D model of the most deteriorated part of the wall is presented in Figure 11. Multiple missing bricks can be identified. Measurements of cavities are enabled. Using the same procedure as in the case of building no. 1, identification of different types of bricks and places which are prone to saturation was conducted. In Figure 12, an exemplary area of the wall is presented where both types of the phenomenon are present. Using both types of digital images (ordinary photograph and image in artificial colours mirroring values of intensity), it is reasonably easy to identify different types of bricks and exposure to moisture. As in building no. 1, each change on the surface (physico-chemical properties) affects the absorption and dispersion of the laser beam, which translates into the obtained value of intensity. Stains created by rainwater (large blue area in artificial colours) are clearly visible.

4. Conclusions

The study involved reviewing the possible applications of the TLS technique for assessment of heritage brickwork buildings. The proposed approach was tested on two different brickwork buildings located in Poland. The identification of different types of bricks, missing bricks, missing mortar and presence of moisture was proven effective. In the authors’ opinion, it is possible to use an ordinary TLS apparatus for quick and efficient assessment of the technical state of heritage brickwork buildings and structures. Further research programmes should cover the development of dedicated software for TLS to enable precise detection of moisture, lime stains and volume of missing bricks. Different types of TLS should be used for the type of scans in question. The use of different models of TLS (e.g., using different wavelengths of the laser beam) may result in achieving different values of intensity while scanning the same brickwork structure under constant conditions. The possible changes in the construction of specific TLS apparatuses should be explored. Development of new types of TLS is needed to ensure effective data collection, apart from basic 3D object location.