Assessment the Contribution of the Major Quality Influencing Factors on the Measurements of TLS Scan

doi:10.21203/rs.3.rs-4977482/v1

Download PDF

Research Article

Assessment the Contribution of the Major Quality Influencing Factors on the Measurements of TLS Scan

https://doi.org/10.21203/rs.3.rs-4977482/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Surface qualities, range detectors, and measuring travel time are all major quality-influencing factors that affect the position accuracy of the measured point clouds by the terrestrial laser scanner (TLS). We conducted experimental tests using ScanStation2 TLS to investigate the accuracy of the scanned point clouds at various incident angles and ranges, and then examined the influence of different scanned surfaces on roughness and reflectivity. In this study, we selected four distinct surface properties from various materials (glass, steel, wood, ekoplast, and adhesive total station (TS) target), and painted three of these materials in five different colors to investigate the influence of colored surfaces on the reflected measured point clouds. During the experiment, approximately 250 scans were recorded, as the chosen materials were scanned at six scan angles (0, 15, 30, 45, 60, and 75) and ranges of 5m, 20m, 40m, and 60m. The experiment's findings show that, at various incident angles, smooth surfaces have a greater impact on the accuracy of the measured 3D points than do rough surfaces. Conversely, the total RMSEs of the red and black colors were greater than those of the blue, green, and white colors. At 0˚ incident angle, the TS target reflects approximately 20 cm closer to the TLS than the other materials; this difference decreases as the scan angle increases. In comparison to the other materials, the difference becomes about 2 mm at a 75˚ incidence angle. With the exception of the 30˚ scan angle of wood material, the maximum RMSE of rough materials is less than 1 cm, while the highest RMSE for smooth surfaces at 45˚ glass material is 4 cm. Moreover, the intensity of different materials varies significantly. For example, smooth materials like steel and glass have varying degrees of accuracy because of their respective properties. We have created a best-fit patch for all the scanned points to detect their deviation and characterize a suitable correction method. Due to the huge number of point clouds that resulted from those hundreds of scans at different conditions, it is very difficult and complex to directly apply a point position correction for all those complicated scanning conditions. Therefore, in this study, a comprehensive and intensive Python programming code was developed to correct a large number of point cloud positions within a standard processing time. This, in turn, contributes to the process's time and cost savings. Interestingly, this developed code is a novel procedure for correcting TLS point clouds at different measurement conditions, so it will be a good suggestion to add it to the Cyclone software.

TLS

point clouds quality

major quality factors

surface reflectivity

scan angle

scan range

best fit patch

error correction

program coding

In fact, laser scanner systems use two-way travel times (TOF) (time of flight) to calculate the distance (ρ) between the scanned objects and the scanner. Placing targets on the surface of the items to be scanned is not necessary; reflector-less scanner systems are typically preferred in the scanning process. Therefore, the range is dependent on how much of the object's sufficient returned signal reaches the scanner's photodetector. The material's characteristics, therefore, influence how much of the reflecting signal reaches the scanner range finder.

There are numerous approaches, ranging from the conventional way to contemporary methods, to produce an accurate 3D model of the items. Image-based and laser-based techniques, which use a camera and sensor, are the most popular and recent methods (T. On Chan & Lichti, 2012). Terrestrial laser scanning (TLS) is the most modern and precise method in this sector.

Remarkably, a wide range of parameters, including material qualities, measurement geometry, meteorological conditions, and instrument effects, all had an impact on the recorded point clouds (S. Soudarissanane, 2016). The scanners' appealing features include high spatial density data, high measurement data, and the reflector-less natural scanning technology's accuracy. However, the instrument's geometry (Tan et al., 2018), the oscillating mirror problem (Bae & Lichti, 2007), beam divergence (Sylvie Soudarissanane et al., 2009), scan angle (Sylvie Soudarissanane & Ree, 2007), the calibration process (Abbas et al., 2014) and (Ting On Chan et al., 2015), and the type of instrument based on the manufacturer, such as Topcon, Faro, Leica, and Trimble.

However, one of the main variables influencing the quality of the observed point clouds, along with incidence angle and distance, is scanning geometry (Amer et al., 2018; (Derek D. Lichti & Harvey, 2002; Sylvie Soudarissanane & Ree, 2007). While there were many other problems during the processing of data in the registration and georefrecessing process, targets method process, and some software obstructions (Bae & Belton, 2012), (Murtiyoso & Grussenmeyer, 2018), (Abdurrahman Farsat1* et al., 2014), (Abbas et al., 2014), (Date et al., 2019), (Steinvall, 2007), and (Jr et al., 2017), Various materials were scanned for various purposes, such as deformation monitoring, monitoring of the bridge structure (Hartmann & Alkhatib, 2023), measurement of building facades (Balzani, n.d.), and Helios-Rybicka & Förstner, 1986). Obviously, scanning these types of materials provides different qualities and accuracies due to color effects, surface quality, and material types (Huang et al., 2023), (Julin et al., 2020), and (Huo et al., 2023).

This study focuses on performing various scans with specific parameters for various materials under identical weather conditions. Its goal is to investigate how scanning various materials' smooth and rough surfaces and colors affects the precision and clarity of the measured point clouds at various incident angles and distances from the TLS to objects. Next, we will develop a best-fitting patch for the scan data, write a coding program, and create an application to reduce the noise caused by these impacts on the material's surface.

2.1 Materials

Surface roughness and reflectivity are two crucial characteristics that could significantly affect the precision and quality of the measured points for reflectorless instruments, regardless of the material's construction from natural or artificial materials (Derek D. Lichti & Harvey, 2002).

This research technique involved painting three separate surfaces in five different colors, and selecting and prepping four distinct materials for scanning. They used two primary criteria for their selection: first, the most usable material found in man-made objects, including towers or buildings. Second, the material's surface roughness and reflectivity are noticeably different. Glass, steel, wood, and Ekoplast were therefore the materials used, while the painted materials were Astropol, gypsumboard, and alicobon. Man-made construction mostly employs these materials, each with a significantly varying surface roughness.

Before initiating the scanning process, we scrutinized the characteristics of each selected material, gathered all essential information, and established the sample's dimensions, which are as follows: We use glass with a thickness of 6 mm and dimensions of 80 cm by 80 cm as a mackle to reflect the laser signal to the scanners and minimize noise. In termsConstruction projects, such as building and bridge construction, frequently utilize steel as their maThe steel sample measures 80 cm by 80 cm and has a 1.8 mm thickness.

Additionally, bookcases, lockers, doors, and structures primarily employ wood in their construction. Its example was made entirely of wood without any painting or covering, and it had the same dimensions (80 cm by 80 cm) and a thickness of only 8 mm. Lastly, plasterers typically use ekoplast to separate and release sound, heat, and cold from a building's façade. To create products like gypsum, stucco, and parget, combine this substance with water. Many internal components filter heat and cold from the outside into the building and vice versa. Therefore, we have constructed an ectoplasm sample for this study, measuring 20 mm in thickness and 80 cm by 80 cm in size.

The painted materials state that alicobon is one of the materials most frequently used in construction, including for writing tableaus and decorating buildings. Therefore, we must consider the reflectivity and roughness properties of the material, depending on the colors chosen to reflect the laser beam's signal. This material is 80 cm by 80 cm and has a thickness of only 18 mm. We painted the initial face, Fig. 1, in all five colors. We divided it into 25 squares, each measuring 16 by 16 cm. Every row and column contained every color. We applied four colors to the other face, then divided it into four 40 x 40 cm squares.

Gypsum board: This is another substance that is frequently used for plastering the interior walls of structures. Based on the colors, it is necessary to take into account the material's reflectivity and roughness properties. This material is in between rough and smooth based on the diameter of the laser beam footprint. Measuring 80 cm by 80 cm, it is barely 10 mm thick. Painted in green and red, the first face, Fig. 1, was divided into two rectangles, each measuring 40 by 80 cm. The other side was painted the same size as the first face in shades of blue and red.

Astropol: Another substance frequently used for plastering the outside walls of buildings is Astropol. Therefore, it is necessary to take into account the material's reflectivity and roughness. Bidirectional Reflectance Distribution Function (BRDF) and laser beam diameter allow materials to be classified as rough in terms of roughness. It is 80 cm by 80 cm and 50 mm thick. Figure 1 shows a single face painted in black and white that is separated into two rectangles, each measuring 40 by 80 centimeters.

However, based on the size of the footpring area of the laser beam and BRDF, the materials glass, gteel, and alicobon are regarded as smooth surfaces, but the materials wood, ekoplast, gypsumboard, and astropol are seen as rough surfaces based on the beam's footprint and BRDF. Figure 1 displays all of the prepared and chosen material samples used in this study.

In order to investigate the impact of smooth and highly reflecting surfaces on the quality of observed points at varying scan angles, five or three total station targets were fixed on the sample of each material that was chosen. The kind of TS target is a 4 × 4 cm adhesive sheet target, and it has a thickness of only around 1 mm. These targets have the maximum amount of reflectivity and are perfectly smooth..

2.2 Methodology

This study's approach was carried out in three primary steps: In order to comply with the specifications for an accurate scanning technique, the first step is to choose and prepare material samples that satisfy the necessary dimensions. It also involves building the necessary stand to hold the material samples correctly. The second stage involves configuring the TLS and establishing the distance between the material stand and the scanning station. Lastly, carry out the procedure of scanning the material sample at various scan angles as the third phase. The technique flowchart used to accomplish the goals of this research is shown in Fig. 2 below.

It is crucial to note that the creation of a stand is essential for achieving accurate and high-quality material sample scanning. This is because the range between the material sample and the scanner instrument must be fixed as an observed target, the scanning angle must be controlled, and the material sample must remain vertical throughout the scanning process. Wood was used to make this platform because it was inexpensive and lightweight, making it easy to move from one location to another. It has two distinct sections in terms of design and structure: The first component (top), a frame with dimensions of 1 x 0.8 m, is used to place and secure a sample of the material inside. Interestingly, this frame is made to be a detachable component from the body of the construction, making it simple and quick to remove, alter, and then reattach a sample of a different material. As seen in Fig. 3, the second component (lower) is a circular stand constructed of wood. This component is 50 centimeters above the ground. To create the verticality of the first frame, this component of the construction has two circular bases, one up and one down. At the center of these 2 circular bases are 2 central wholes that lie on a single vertical line, as seen in Fig. 3. The Topcon total station instrument's centering laser beam was utilized to regulate this verticality. Additionally, a sharp nail next to the edge of the upper circular portion controlled the scan angles.

2.3 Scanning Materials

The samples of all prepared, chosen materials were scanned at 6 distinct scan angles (0˚; 15˚; 30˚; 45˚; 60˚; and 75˚), and there were four different distances of approximately 5 m, 20 m, 40 m, and 60 m between the instrument and the material's stand. This indicates that 24 scans were carried out for each item. Figure 4 displays some scanned data examples.

According to Table 1 below, the ScanStation2 TLS instrument's specifications were followed when using it for scanning. Moreover, the scanned data was directly sent into the linked PC via an Ethernet cable and stored and analyzed using Leica licensed cyclone software.

Table 1

ScanStation2 laser scanner specifications
Names	Specifications
Name of Instrument	ScanStation2; Leica Geo-system; class-3R
FoV	V. angles 270˚ and H angles 360˚
Two Scanning windows	1st, w. -45˚ to 32˚ and 2nd w. 23.5˚ to 90˚
Beam divergence	After 20m range
Beam diameter	2 mm
Scanning model	Pulse-based method
Accuracy of a single point	Position-6mm and 4mm-distance
Maximum Range	300 m
Scanning rate	Up to 50000 Pts./Sec.

Interestingly, five adhesive TS targets were positioned at the center and corners of the surface of the scanned material sample to investigate how smooth and highly reflective surfaces affected the quality of the observed point clouds at various scan angles.

3.1 Results

Because of its high degree of object scanning precision, TLS is an instrument that has been used recently in many different industries, such as building, oil tank calibration, documentation, monitoring, and many more. Like other instruments, it is not immune to mistakes that arise in measured point clouds. Therefore, four different materials (in terms of reflectivity and roughness) were scanned at varying scan angles and ranges in order to examine the extent of their effects on the quality and accuracy of the measured point clouds. In contrast, three different materials and five different colors are used to differentiate between the reflectivity and roughness of the measured point clouds at different incident angles and ranges in order to assess their quality.

3.1.1 Material Effects

As previously noted, samples of selected materials, including glass, steel, wood, and ekoplast, were used to conduct experimental tests. It is significant to note that the amount of reflected point clouds is observed to vary with the temperature, roughness, and properties of individual materials. For instance, because of the substantial differences in surface roughness between the two materials, there were fewer reflected point clouds from the glass material than from the wood at the same incident angle and distance. Furthermore, as Fig. 5 illustrates, it is evident that when the incident angle increases, the number of reflected point clouds from materials such as steel and glass falls significantly, while the number from materials such as wood and ekoplast declines gradually.

3.1.2 The Sun Shine Effects on Smooth Surfaces

It's interesting to note that, as Fig. 7 illustrates, there was a serious issue when the deviation from the fitted patch was determined. A complete white region was displayed in the upper-left corner of the image plan, signifying, according to science, an extremely high-intensity reflection for steel material. Similarly, for glass material, the same region was displayed in the middle right. This is the result of the sunlight striking the steel and glass surfaces at precisely the right angle from the scanner. This causes the scanned laser beam to become more intense, returning the reflected point clouds to the scanner sooner than they would have otherwise for the steel material. In contrast, for the glass material, the point clouds return behind the glass because the sun's rays create a lens behind the glass surface. The glass sample is scanned at a 30˚ incident angle in the 20m and 40m ranges in Fig. 6, while the steel sample is scanned at a zero degree incident angle in the 5m and 20m ranges. As can be observed, at a distance of between 5 and 40 meters from the scanner to the item on smooth surfaces, the sunlight effect is computed as a deviation value, or around 2 to 5 cm.

Table 2: Problems on the smooth surface materials (glass and steel) when the sunshine coincides with the incident angle

Glass-20m-30d	Glass-40m-30d	Steel-5m-0d	Steel-20m-0d
0.024	0.036	0.039	0.03
0.029	0.033	0.054	0.038
0.015	0.031	0.028	0.027
0.018	0.021	0.033	0.037
0.024	0.028	0.028	0.029

3.1.3 Coloured smooth surface effects on the quality of the point clouds

Based on the scanning geometry and distances between the scanner and the object, the color of the material was affected by the reflectivity and roughness of the material. According to the experimental results, these colors that were used have different effects on the accuracy and quality of the points. For the perfect smooth materials, the effect of the color highly appeared on the quality of the point clouds. For example, black smooth objects at the 60-degree incident angle did not reflect any points, while for the rough materials with the same incident angle and color, some points were reflected to the scanner. On the other hand, when the scan angle is equal to 0˚ (indicating the object is directly in front of the scanner) and the laser beam hits the object approximately perpendicularly, the reflected points in all of the colors and all of the surface properties (rough, middle, and smooth) of the of the materials are approximately equal to each other. Figures 8 and 9 below illustrate the results.

3.1.4 Coloured Rough Surface Effects

The above-mentioned tests are performed for smooth materials according to the diameter of the footprint area of the laser beam. On the other hand, two other different materials are used for testing these colors, which are highly affected by the incident angle when the material is smooth. These two materials are rough based on the diameter of the laser beam and BRDF. Figure 10 illustrates the experimental tests on the rough materials based on different colors and angles.

3.1.5 The Effect of Distance on the Quality Point Clouds

Another factor that affects the quality of the point clouds is the distance between the scanner and the scanned object. Before starting to scan the objects, the sample spacing between point clouds can be changed in some of the scanners. For example, the distance between two adjacent points is equal to 1 cm at two different distances, such as 5 m and 20 m. By using this method, you can control and keep track of the number of point clouds per unit area. But many of the factors remain on the accuracy and quality of the points, such as the periodical time of the scanned area increasing when the range increases between the scanners and object; the other factor is the beam divergence increasing when the range increases, which means the footprint area per point cloud increases based on the range. This means that the amount of returned laser beam power decreases and the intensity decreases with increasing range. The other factor is during the travel of the laser beam in the atmosphere; for longer distances and times, the effect of the refraction and dust on the accuracy and quality of the points increases compared to short distances.

It is noteworthy that the number of reflected point clouds varies with respect to each material's quality, heating, roughness, incident angle, and range. For instance, because both materials' surfaces differed significantly in terms of roughness and reflectivity, there were fewer point clouds reflected from glass material at the same incident angle and distance than from wood. Furthermore, as Fig. 11 illustrates, it is evident that as the incident angle increases, the quantity of reflected point clouds from materials with smooth surfaces (such as steel and glass) and materials with rough surfaces (such as wood and ectoplasm) gradually decreases. On the other hand, the number of point clouds decreases with increasing range for the same material and incident angle (see Fig. 12).

By increasing the incident angle and range, the number of points per patch decreases. Table 3 below shows an example of decreasing point clouds based on increasing incident angle and range. While the sample spacing for all scan ranges is similar, it is 1 cm. For example, for 5 m, it is considered 1 cm per 5 m, and for the 60 m range, it’s considered 1 cm per 60 m.

Table 3

The effect of incident angle and distance on the density of point clouds
Material	Incident angles	Points used to patch	Material	Ranges	No. of Points
Wood-5m	0˚	6065	Wood-30˚ incident angle	5m	5085
	15˚	5829		20m	5119
	30˚	5085		40m	4900
	45˚	3790		60m	4823
	60˚	2632
	75˚	1132

3.1.6 The Effect of Total Station Targets

Five sticky total station targets were also affixed to the material's surface in various locations. One point cloud is chosen for each target in order to determine the deviation value from the fitted patch. Therefore, it is evident from Table 4 and Fig. 13 that the deviation values of the points decrease in all ranges from zero to sixty degrees of impact angle. The deviation value ranged from 12 to 20 cm when the incident angle was at zero degrees. This difference at zero degrees was observed when the range changed from 5 to 60 meters. For all materials, the deviation value decreased to less than 4 cm when the incident angle reached 60˚, and it was less than 1 cm at 75˚. This indicates that when the incident angle reaches 60° or higher, the scanner is unable to detect the impact of the material attributes. The deviation of these targets on materials with rough surfaces is smaller than that of materials with smooth surfaces. On the other hand, for the same materials with different ranges, the deviation from the best-fit patch will decrease with increasing range. It clearly appeared in the 40- and 60-meter ranges because the size of the footprint area of the point cloud increased and the intensity decreased. That affected detecting the exact measurement of the property of the object's surface.

Table 4

The effect of (5m, 20m, 40m, and 60m) distance and (0˚, and 60˚) incident angle on the adhesive total station targets

Degrees	5m	20m	40m	60m
0d-Glass	0.1932	0.1942	0.1476	0.1262
0d-Steel	0.1888	0.1956	0.1476	0.115
0d-Wood	0.157	0.1864	0.1662	0.1276
0d-Ekplast	0.1772	0.1864	0.1238	0.1190

Degrees	5m	20m	40m	60m
60d-Glass	0.0354	0.0368	0.038	0.0236
60d-Steel	0.0152	0.0306	0.0302	0.0126
60d-Wood	0.0264	0.0384	0.0244	0.0116
60d-Ekoplast	0.0122	0.0340	0.0144	0.0126

3.2 Overall RMSEs of Materials

Table 5 presents the overall root mean square errors (RMSEs) for all of the selected scan materials used in this study. The smooth materials (steel and glass) had the highest RMSE, while the rough materials (wood and ekoplast) had the lowest RMSE. When the smooth glass surface is incident at a 45-degree angle, the greatest RMSE is 4 cm, and the rough ekoplast material at the 75-degree incident showed the lowest RMSE, which is 2.6 mm. The smooth materials have a minimum RMSE of 0.6 cm and a maximum RMSE of 4 cm. On the other hand, the rough materials have a minimum RMSE of 0.26 cm and a maximum RMSE of 1.46 cm. It means the quality of the measured point clouds on the rough surfaces is approximately two times higher than the quality of the smooth surfaces.

Table 5

RMSEs of the materials' surface at the selected points
RMSE	0d	15d	30d	45d	60d	75d
Glass (cm)	1.33	1.80	3.48	4.03	2.18	1.47
Steel (cm)	1.09	0.81	0.88	1.17	0.89	0.60
Wood (cm)	0.91	0.0088	0.0146	0.0083	0.0046	0.0055
Ekoplast (cm)	0.90	0.63	0.89	0.70	0.47	0.26

Additionally, Fig. 14 demonstrates that for all materials, which are roughly equal, the total departure from the best-fit patch is at a zero-degree incident angle and that this difference grows with increasing incident angle.

The glass material's almost flawless smoothness had a significant impact on the accuracy of the reflected point cloud data. The clarity of the reflected point clouds was unaffected by the practically flawless roughness of the Ekoplast material. The primary reason for this distinction between smooth and rough surfaces is that the presence of micro-facet components on rough surfaces influences the quantity and strength of the signal returned to the scanner at higher incident angles. The impact of both smooth and rough surfaces on the accuracy of point clouds is depicted in Fig. 14.

Basically, for the rough materials, the same problem can be seen in the reflectivity of the materials for different colors that were used for this purpose. But the variation for rough materials is less than for smooth materials; this might be due to the laser signal hitting the micro-facet parts of the rough material reflected laser beam at zero degree when the scanned object is at a non-zero degree incident angle. It means that these micro-facets on the surface of the rough materials will decrease the effect of the incident angle.

Based on the selected point clouds, the overall RMSEs were calculated for all scans, colors, and incident angles. Table 6 and Table 7 show the overall RMSEs of the color effects at different incident angles for both smooth and rough materials, respectively. The red color has the highest RMSE in both smooth and rough objects; it is about 0.013 to 0.020 m, while the black color, which varies from about 0.01 m to 0.02 m for smooth and rough materials, also records the maximum value on the smooth surface. In terms of white, blue, and green colors, they are approximately equal, and the overall RMSE varies from 0.005 m to 0.01 m at all of the incident angles and smooth and rough objects. On the other hand, the overall RMSEs in all cases decrease when the incident angle increases because, when the incident angle increases, the scanner cannot classify and exactly reflect to find the differences in reflectivity of the materials and colors. So, it reflected the laser beam in the same manner for all of the materials and colors.

3.3 Discussion

Due to all of these effects that were mentioned, the position and elevation quality and accuracy of a single point cloud differ from one point to another on the same surface. For the above position and elevation corrections and all of the correction steps, an app was created, and code was written for the program using the Python algorithm. The most important thing in writing this program is that it must be able to read the import data, calculate the distance, incident angle, and deviation from the patch, then perform the corrections in a sequence procedure step by step according to the mathematical procedures, and finally save the corrected data as a separate file.

3.3.1 Applying Correction for Different Surface Material Properties

A correction procedure was performed for four different materials (glass, steel, wood, and ectoplasm) at four different distances (5m, 20m, 40m, and 60m) and six different incident angles (0˚, 15˚, 30˚, 45˚, 60˚, and 75˚). It means 96 scans and corrections were applied. Due to the large number of scan data points for each range, the author selected a material at one different incident angle. Because the range affected the point clouds less than the incident angle and surface properties.

Four different cases appeared on the surface of materials. Firstly, a portion of the laser signal penetrates the glass materials and hits the stand surface, then returns to the scanner. Secondly, five adhesive total station targets were fixed with perfect reflectivity. Thirdly, the surface of the material itself. Finally, in some scans, the sunshine hits the smooth surface materials and causes a problem. All of these problems appeared in the scanned data. The created coding program will solve all of these effects on the point clouds. Below Fig. 15 shows the correction for zero degree incident angle for glass material, 30˚ incident angle for steel material, 45˚ for wood material, and 60˚ for ectoplasm material. The maximum remaining error in the scanned data after correction for all cases is less than 1.5 mm, which is practically ignored.

3.3.2 Apply Correction for Colour Surface Materials

We applied the correction to the smooth surface based on the best-fit patch for each scan. The correction result is shown below Fig. 16 for 5 m ranges. At a 5 m range and a zero degree incident angle, the data fluctuates between − 0.01 m and 0.015 m, and we applied the correction precisely to every point that diverged from the best-fit patch. Conversely, as the incident angle rises to 60 degrees, the overall deviation from the patch reduces to approximately 5 to 10 mm, and we apply a correction to all the measured point clouds. For the entire scanned data, the deviation after correction is less than 1 mm.

Next, we applied the correction to the rough surface using the best-fit patch for each scan. The correction result is shown below Fig. 16 for 5 m ranges. At a 5 m range and a zero degree incident angle, the data varies by approximately − 0.005 m to 0.005 m. This variation is smaller than the smooth colored surface due to the micro-facets on the material's surface. The correction was applied precisely to all the point clouds that deviated from the best-fit patch. On the other hand, when the incident angle increases and reaches 60 degrees, the overall deviation from the patch increases to approximately − 10 mm to 10 mm due to the same noises in the data, and a correction was also applied for all the measured point clouds. The result of the deviation after correction is less than 1 mm for the entire scanned data from zero to 60º incident angle.

TLS is an instrument utilized in many different sectors that demand a millimeter level of accuracy due to its high performance level of accuracy in the production process. Examples of these industries include oil tank calibration, monitoring, and the recording of extremely expensive heritages. Nevertheless, TLS is not immune to mistakes that arise during the data capture process when scanning objects. Three primary aspects determine the accuracy of the measurement data: the scanner's ability to measure the laser beam's time of flight, the characteristics of the materials it scans, and the scanner's range finder. The practical experiments conducted in this research have yielded important results regarding sunshine, surface qualities, colors, and scan angle—all of which should be considered when aiming for extremely high TLS object scanning performance. and resolving the issues using the best-fit patch that was made, as well as developing software to address the issues with position and elevation.

First off, materials with a rough surface have a higher reflectivity than those with a smooth surface because of the micro-faced portions on their surfaces. Rough surfaces of wood and Ekoplast do not record high deviation values from 0˚ degree to 75˚ degrees of scan angle; in contrast, smooth surfaces of steel and glass did record high deviation values at 0˚ to 75˚. With the exception of wood material at 30˚, the maximum RMSE for rough materials is less than 1 cm, whereas the maximum RMSE of glass smooth surface material at 45˚ is 4 cm.

Second, the degree of accuracy in scanning 3D points varies depending on the material due to differences in its roughness, transmission, absorption, and reflection of the reflected laser signal. Because of this, the point clouds' accuracy varied depending on the type of substance. For example, the minimal RMSE in glass material is 1.47 cm, whereas in steel material, it is due to their varied characteristics. Conversely, the steel material's highest RMSE is 1.17 cm.

Thirdly, a significant issue with the accuracy and quality of the single-point measuring on smooth surfaces for steel material arises when the scan angle aligns with the angle at which the sun shines. This also applies to glass materials. As that occurred, the steel material's incidence angle was zero degrees. and, for glass material, a 45º incidence angle.

However, the scanner is unable to identify the characteristics of the materials when the incident angle approaches 60° or greater. All of the materials used in the experiment had reflectance and intensity of reflected point clouds that were near one another. As demonstrated for the adhesive total station sheet targets at incident angles of 60 and 75 degrees, it can be confirmed by Ameer et al. (2018), who found that an increase in the projection angle employed causes a direct drop in scanning accuracy.

Then, the intensity of the colors will affect the number of returned point clouds and the overall RMSE of the scanned surface. When the intensity increases, the reflectivity of the returned signal increases, and the overall RMSE decreases.

The quality and intensity of the returning laser beam had an impact on the range's ability to measure point clouds accurately; this effect was more pronounced at a zero-degree incident angle than at non-zero angles. For adhesive total station targets, for instance, the departure from the best fitting patch at the 0˚ and 5m range is approximately 20 cm, whereas the same incidence angle and 60m range show a deviation of about 12 cm. Additionally, as the range is increased, the periodic time required to scan an object's identical surface increases.

A program was created to address the majority of the noise in the measured point clouds based on all of the issues and conclusions that were found. On glass smooth surface materials, the greatest number of residual errors in a single point cloud under the worst conditions is 1.5 mm after correction, although the majority of point cloud position and elevation errors are zero after correction.

To sum up, the program's work on the location and elevation of the point clouds is crucial. The software can be used for both single scans and single point clouds, depending on whatever best-fit patch was made for the material's surface. While the majority of other academics focused on enhancing the scanned point clouds' texture, intensity, and geometry correction,.

The authors plan to use this coding program as an icon in the upcoming version of the Leica Cyclone software to lessen the impact of incident angle, range, and surface material properties on the quality of scanned materials. This recommendation will be made to the Hexagon Leica Geo-Systems company for future research. It is advised to alter the coding program that uses multiple scans of data after the registration phase in order to lessen the impact of errors resulting from the georeferencing and registration processes.

Author Contribution

We (Bakhtyar Ahmed Mala and Dleen Muhammed Saleh) confirm that the manuscript has been read and approved by all named authors. We also confirm that each author has made the same contribution to the paper. We further confirm that the order of authors listed in the manuscript has been approved by all authors

Acknowledgement

This article has been done in partial fulfillment a Ph.D degree in civil engineering/photogrammetry at Salahaddin University-Erbil. Special thanks are directed to the Civil Engineering Department and Geomatics (surveying) staff.

Funding Declaration

There is no funding Declaration for this paper.

laser scanner. Measurement: Journal of the International Measurement Confederation, 52(1), 111–123. https://doi.org/10.1016/j.measurement.2014.03.009
Abdurrahman Farsat1*, Martin J. Smith, & Nikkolas Kokkas. (2014). The importance of designing targets for improving the accuracy of Mobile Laser Scanning data. Conference 2nd -5th September at Aberstwyth University. Available at Conference Booklet, Technical Section Lidar. P.P 86-92, May.
Amer, H. A., Shaker, I. F., Abdel-gawad, A. K., Ragab, A., & Mogahed, Y. (2018). Accuracy Assessment of Laser Scanner under Different Projections Angles. 428–438.
Bae, K., & Belton, D. (2012). A Framework for Position Uncertainty of Unorganised Three-Dimensional Point Clouds from Near-Monostatic Laser Scanners Using Covariance Analysis. May.
Bae, K., & Lichti, D. D. (2007). On-site self-calibration using planar features for terrestrial laser scanners. ISPRS Workshop on Laser Scanning, 36(1), 14–19.
Balzani, V. (n.d.). Electron Transfer in Chemistry ®WILEY~VCH Weinheim • New York • Chichester Brisbane • Singapore • Toronto.
Chan, T. On, & Lichti, D. D. (2012). Cylinder-Based Self-Calibration of a Panoramic Terrestrial Laser Scanner. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXXIX-B5(September 2012), 169–174. https://doi.org/10.5194/isprsarchives-xxxix-b5-169-2012
Chan, Ting On, Lichti, D. D., & Belton, D. (2015). A rigorous cylinder-based self-calibration approach for terrestrial laser scanners. ISPRS Journal of Photogrammetry and Remote Sensing, 99(June 2018), 84–99. https://doi.org/10.1016/j.isprsjprs.2014.11.003
Date, H., Wakisaka, E., Moribe, Y., Kanai, S., Vi, C., & Vi, W. G. (2019). TLS POINT CLOUD REGISTRATION BASED ON ICP ALGORITHM USING POINT QUALITY. XLII(June), 10–14.
Gordon, S., Lichti, D. D., Stewart, M. P., & Tsakiri, M. (2000). <title>Metric performance of a high-resolution laser scanner</title>. Videometrics and Optical Methods for 3D Shape Measurement, 4309(December 2000), 174–184. https://doi.org/10.1117/12.410872
Hartmann, J., & Alkhatib, H. (2023). Uncertainty Modelling of Laser Scanning Point Clouds Using Machine-Learning Methods. Remote Sensing, 15(9). https://doi.org/10.3390/rs15092349
Helios-Rybicka, E., & Förstner, U. (1986). Effect of Oxyhydrate Coatings on the Binding Energy of Metals by Clay Minerals. Sediments and Water Interactions, September, 381–385. https://doi.org/10.1007/978-1-4612-4932-0_32
Huang, X., Zhang, Y., Ni, B., Li, T., Chen, K., & Zhang, W. (2023). Boosting Point Clouds Rendering via Radiance Mapping. Proceedings of the AAAI Conference on Artificial Intelligence, 37(1), 953–961. https://doi.org/10.1609/aaai.v37i1.25175
Huo, L., Liu, Y., Yang, Y., Zhuang, Z., & Sun, M. (2023). Review: Research on product surface quality inspection technology based on 3D point cloud. Advances in Mechanical Engineering, 15(3), 1–17. https://doi.org/10.1177/16878132231159523
Jr, L. G., Bordin, F., Veronez, M. R., Marson, F., & Scaioni, M. (2017). A new approach to minimize border effect for terrestrial laser scanning A NEW APPROACH TO MINIMIZE BORDER EFFECT. July. https://doi.org/10.1109/IGARSS.2017.8127201
Julin, A., Kurkela, M., Rantanen, T., Virtanen, J. P., Maksimainen, M., Kukko, A., Kaartinen, H., Vaaja, M. T., Hyyppä, J., & Hyyppä, H. (2020). Evaluating the quality of TLS point cloud colorization. In Remote Sensing (Vol. 12, Issue 17). https://doi.org/10.3390/RS12172748
Lichti, D. D., Stewart, M. P., Tsakiri, M., & Snow, A. J. (2000). Calibration and testing of a terrestrial laser scanner. International Archives of Photogrammetry and Remote Sensing, 33(B5/2; PART 5), 485–492.
Lichti, Derek D., & Harvey, B. . (2002). the Effects of Reflecting Surface Material Properties on Time-of-Flight Laser Scanner Measurements. Geospatial Theory, Processing and Applications, 2001.
Murtiyoso, A., & Grussenmeyer, P. (2018). Comparison and assessment of 3d registration and georeferencing approaches of point clouds in the case of exterior and interior heritage building recording. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives, 42(2), 745–751. https://doi.org/10.5194/isprs-archives-XLII-2-745-2018
Soudarissanane, S. (2016). The Geometry of Terrestrial Laser Scanning: Identification of Errors, Modelling and Mitigation of Scanning Geometry. https://repository.tudelft.nl/islandora/object/uuid%3Ab7ae0bd3-23b8-4a8a-9b7d-5e494ebb54e5
Soudarissanane, Sylvie, Lindenbergh, R., Menenti, M., & Teunissen, P. (2009). Incidence Angle Influence on the Quality of Terrestrial Laser Scanning Points. Iaprs, XXXVIII, 183–188.
Soudarissanane, Sylvie, & Ree, J. Van. (2007). Error budget of terrestrial laser scanning : influence of the incidence angle on the scan quality. May 2014. https://doi.org/10.13140/RG.2.1.1877.6404
Steinvall, O. (2007). Influence of laser radar sensor parameters on range-measurement and shape-fitting uncertainties. Optical Engineering, 46(10), 106201. https://doi.org/10.1117/1.2789654
Tan, K., Zhang, W., Shen, F., & Cheng, X. (2018). Investigation of TLS Intensity Data and Distance Measurement Errors from Target Specular Reflections. https://doi.org/10.3390/rs10071077

Tables 6 and 7 are available in the Supplementary Files section.

No competing interests reported.

Tables6and7.docx

Download PDF

Reviewers agreed at journal
10 Sep, 2024
Reviewers invited by journal
09 Sep, 2024
Editor assigned by journal
27 Aug, 2024
Submission checks completed at journal
27 Aug, 2024
First submitted to journal
26 Aug, 2024

You are reading this latest preprint version

Assessment the Contribution of the Major Quality Influencing Factors on the Measurements of TLS Scan

Status:

Version 1

Abstract

Figures

1 Introduction

2 Materials and Methodology

2.1 Materials

2.2 Methodology

2.3 Scanning Materials

3 Result and Discussion

3.1 Results

3.1.1 Material Effects

3.1.2 The Sun Shine Effects on Smooth Surfaces

3.1.3 Coloured smooth surface effects on the quality of the point clouds

3.1.4 Coloured Rough Surface Effects

3.1.5 The Effect of Distance on the Quality Point Clouds

3.1.6 The Effect of Total Station Targets

3.2 Overall RMSEs of Materials

3.3 Discussion

3.3.1 Applying Correction for Different Surface Material Properties

3.3.2 Apply Correction for Colour Surface Materials

4 Conclusion

Declarations

Author Contribution

Acknowledgement

References

Tables 6 and 7

Additional Declarations

Supplementary Files

Status:

Version 1