THE MULTITEMPORAL PHOTOGRAMMETRIC DATA IN ARCHAEOLOGICAL AND ARCHITECTURAL RESEARCH IN THE ROYAL CASTLE IN WARSAW-FIRST RESULTS

Modern measurement technologies are commonly applied not only to monitor Cultural Heritage objects; they are also applied during archaeological excavation works, when it is important to quickly perform measurements. The paper presents multitemporal integration of different image-based (UAV, close-range digital images) and range-based technologies (Terrestrial Laser Scanning), as well as data acquired in different periods, during archaeological works performed at the Royal Castle in Warsaw, especially for the Justice Court Tower. Measurements were performed in several periods, during deep archaeological excavation works. Due to the limited access to excavations different measurement technologies were applied which allowed to perform measurements within short time. As a results, the DSM (Digital Surface Model). the point cloud and orthoimages were generated. All of the products were stored in the GIS system which will be used for the needs of archaeological and architectural analyses.


INTRODUCTION *
The Justice Court Tower, which is the subject of this study, is the oldest brick building of the Royal Castle in Warsaw, the edifice which is almost entirely reconstructed after the destruction it suffered during The Second World War.The number of relics from different historical periods included in the castle buildings is exiguous, these are primarily foundations and basement rooms.This also applies to the Justice Court Tower, which has been dated back to the 14th century.The original cellar located in the basement of the Tower, in the 17th century was adapted to the city prison.From this period come marks carved in bricks, left on the walls of the basement by prisoners.The selection of the object of research and analysis was dictated by conservation reasons.As a result of excessive moisture in the walls, the signs underwent enormous degradation, which in turn, resulted in slow disappearance of the rites.In order to prevent further degradation, it was necessary to undertake quick preservation actions and to create a protection program for decaying carved signs.For this purpose, it was necessary to plan construction works both on the outside of the cellar walls and inside the basement.The necessity of unveiling the outer walls to the level of the tower's foundation gave the opportunity to examine the object in terms both archaeological and architectural.The research program also includes the conservation through documentation.Chosen isolation methods assumed a permanent cover of the historical elements.It was then the only possibility to inventory at least one of the tower's walls in their entirety-from the foundation to the roof.Research and inventory were considerably stretched over time, which was closely related to the construction work schedule and security * Correasponding author considerations.The works were carried out in a deep trench which forced the protection of its the western wall.The piling method was adopted.Due to the applied method, causing significant vibrations, it was decided to make measurements indicating possible displacement of the tower body.Initially, the work was carried out using mechanical equipment, which was used to remove the layers associated with the reconstruction of the Castle.Their thickness reached a depth of 4 m.At the end of this stage, it was possible to take measurements of the exposed, the brick part of western facade of the Justice Court Tower.The next phase was manual exploration up to the foundation level of the Tower.A layer of about 4 m thickness was examined, as a result the foundation plinth was unveiled in its entirety, the height of which was 3 m.At this moment, archaeological research was completed.The next stage was a full inventory using non-invasion photogrammetric methods.

METHODOLOGY
For archaeological research, it is important to appropriately record, document, and survey artefacts and sites because an accurate and complete digital documentation is a prerequisite for further analysis and interpretation.For that purpose methods of 3D shapes measurements are applied, basing on active and passive techniques (Arif and Essa, 2017;Grussenmeyer and Yasmine, 2004;Hatzopoulos et al., 2017;Markiewicz et al., 2017;Remondino and El-hakim, 2006) Selection of an appropriate measurement techniques determines the way of data acquisition and processing as well as development of documentation.Both, the active (TLS) and passive (imagebased approach) techniques have their advantages and disadvantages (Gonizzi Barsanti et al., 2013) Barsanti et al., 2013) Nowadays, research works concerning the Cultural Heritage and archaeology utilise integration of many techniques, including aerial and UAV (Arif and Essa, 2017;Del Pozo et al., 2017;Grussenmeyer and Yasmine, 2004;Koistinen, 2004;Murtiyoso et al., 2017;Nocerino et al., 2012;Sauerbier and Eisenbeiss, 2010), LIDAR -ALS and TLS (Arif and Essa, 2017;Del Pozo et al., 2017;Drap et al., 2007;Gonizzi Barsanti et al., 2013;Grussenmeyer and Yasmine, 2004;Hatzopoulos et al., 2017;Markiewicz et al., 2017), as well as close-range photogrammetric (Gonizzi Barsanti et al., 2013;Grussenmeyer and Yasmine, 2004;Hatzopoulos et al., 2017;Remondino and El-hakim, 2006).For the purposes of discussed analyses the authors used UAV, terrestrial and low-cost photographs and TLS data were used as reference data.Additionally, they all were processed in one reference system, basing on geodetic measurements.

DATA COLLECTION
The division of the inventory process of the tower walls was necessary due to the phased character of the renovation works and limited, overtime, accessibility of individual parts of the walls.The documentation of the tower's ground part was the easiest to do.This stage of work was limited only by atmospheric conditions and it was made in the first place.However, a significant part of the historical facade is located under the surface of the ground, covered by layers formed over several centuries.It is the part that is adverse affected by the environment in which it lays.The result was a serious moisture in the walls which caused poor conservation conditions created in the interior located in the basement.It was necessary to completely uncover the outer western facade and secure the brick part from further degradation by applying vertical insulation.The revealed part of the walls largely consists of original fourteenth-century elements (Kąsinowski, 1960) and unveiling it gave the opportunity to examine its structure.According the schedule of renovation works the next step was to secure the brick part before uncovering the stone foundation.
The project assumed putting on the walls ever a insulating layers permanently covering the historical facade.It was important to take an inventory of the exposed wall before it was insulated, which had to be done before undertaking further research.The lowest parties, consisted of the stone foundation of the Tower, were uncovered in a much narrower trench than the other elements, which also forced a change of the documentation method.The phased of the actions was forced by the schedule of conservation works and the possibility of adapting the documentation method to the existing conditions during the research is a great convenience and the integrated data can be used not only to create a visualization and virtual reconstruction of the examined object, but above all are the subject of many useful analysis.
Figure 1.The tower with marked heights

The subject of the study
The detailed subject of analyses discussed in this paper was the monumental Justice Court Tower located at the edge of an escarpment.Both, its location and dimensions require particular attention during the positioning of the measuring stations.On the one hand, the high tower (H=14 m) is inaccessible for direct measurements, and the partial projection of its shape is possible, on the other (Fig. 1A).

Performed measurement
For that purpose, in order to inventory a multitemporal archaeological excavation it was decided to apply different photogrammetric techniques, i.e. the close-range s -Z+F 5006h, UAV DJI Phantom 3 Pro, close-range images -Canon 5D Mark II, the low-cost, Olympus C-5050Z sensor and the mobile phone camera included in Samsung Galaxy J3 (Tab.2).Measurements were divided into three stages (Fig. 1B).Works were performed by two teams, i.e. a team of surveyors and photogrammetrists from the Warsaw University of Technology and a team of trained archaeologists from the Royal Castle.
One, uniform and common reference system for all types of data is an important element which allows to integrate data acquired from the UAV, TLS and cameras.Besides, considering the multitemporal inventory, this system must be stable in time.The reference system consists of two elements: the coordinate system and stabilised points of the control.The local system was assumed as the X,Y,Z coordinate system.Coordinates of the control points are the 1-order control for referencing all types of data; the appropriate construction for the control allows to analyse its stability in time.Table .2. The overview of used equipment with defined measurements parameters Measurements were performed using the Leica 1201+ total station which allows to measure angles with 1" and the reflector distance measurement with 1 mm +1.5 ppm.The positioning accuracy for the 1-order control equals to 2 mm.The geodetic control was measured in two epochs (I and II).Displacements of reference points were not noticed, after the analyses and comparison of positions of points in both measurement epochs.
The accuracy of determination of points of the photogrammetric control, not worse than 3 mm was obtained.
Figure 2. The ground control points on tested area.A) The example of the used target, B) The sketch of distributed ground control and check points.Red colour -points, the blue rectangle -tested area.The photogrammetric control check points were also defined (Fig. 2A) around the analysed excavation and its closest surroundings (Fig. 2B).

Data processing
In order to generate the final photogrammetric documentation it was necessary to process and integrate data in the State Geodetic Coordinate System 2000.Agisoft PhotoScan software was used to orientate and process images; to process TLS data Z+F LaserControl and LupoScan software were used.The results of data registration was shown in Table 4.The process of terrestrial laser scanning data orientation was performed in two stages: the relative and the absolute orientation was performed in order to register data in the assumed reference system.Marks dedicated for the Z+F 5006h scanner were used as reference points.
At the first stage images acquired from the low-cost UAV were processed using the Agisoft PhotoScan software.Due to destination of images, nadir and oblique images were separately processed.Approximate elements of the exterior orientation, acquired during the flight, were used as initial data.For nadir and oblique images the calibrated focal value, coordinates of the principal point, radial distortion parameters (k1, k2 and k3) and tangential distortion parameters (p1 and p2) were determined in the self-calibration process.Additionally, it was decided to determine parameters related to the rolling shutter effect for nadir images.
As a reference data, the ground control (6) and check (4) points from the geodetic measurements were used.Close-range image processing was divided into three epochs.
The following parameters were determined for each case in the self-calibration process: the calibrated focal value, coordinates of the principal point, parameters of the radial distortion (k1, k2 and k3) and parameters of the tangential distortion (p1 and p2).

UAV images
The obtained results prove the correctness of the performed data orientation process.Lower accuracies obtained for oblique images resulted from projection of control points under bigger angles what considerably affected the accuracy and uniformity of identification of the mark centre.

Close-range images
The obtained results prove the high orientation accuracy of close-range images in particular epochs.However, it will be necessary to perform the reciprocal orientation of all data in the process of integration of data from different sources, in order to eliminate errors generated during the orientation of images from different measurement epochs.Comparing the results obtained from Olympus, the old digital camera (epoch II) with data from the Samsung Smartphone (epoch III) it should be noticed that higher accuracies are obtained for the Olympus camera.Despite the lower matrix resolution results are considerably better since the camera is equipped with the lens of the better geometric properties.

RESULT OF DATA INTEGRATION -THE COMPLEX INVENTORY
In order to perform comprehensive assessment of conditions of the monumental object, the Justice Court Tower, it was necessary to integrate multitemporal data acquired by different sensors, used in many epochs.In order to evaluate the accuracy of performed orientation of photographs ,TLS data were applied as reference data.The possibility to place geodetic control points was limited by the nature of the researched object and by performed archaeological works.This also caused that numbers of points used for particular epochs were different.However, analyses of point clouds were performed in order to assess the accuracy of particular stages of image processing.
Figure 3. Marked tested areas on the orthoimages from all of the epochs

Analyses of differences between TLS point clouds and photographs in particular epochs
At the initial stage, differences between point clouds from images and TLS point were evaluated for particular epochs.For the needs of comprehensive evaluation of integrated point clouds analyses were performed for 5 control sites, where archaeologists will perform detailed archaeological and architectural analyses (Fig. 3).The objective of such detailed analysis was to test the accuracy of the shape projection which may be achieved for particular sensors and to check whether it meets the requirements of such analyses.
Summarising the results obtained for particular sensors (Appendix 1, Tab. 5) diversification of results may be noticed, which may be considered negligible for archaeologists in the case of large excavations.This leads to the general conclusion that both, a full frame camera, as well as a Smartphone camera may be used.However, the appropriate photograph processing must be considered.

Epoch
Deviations between point clouds < 2.5  In order to determine the parallelism of planes it was necessary to determine the scalar product of normal vectors of planes.
When planes are parallel the coefficient equals to 1.As a result of performed analyses, the coefficient equalled to 0.9996.Therefore the value of twist of planes and the maximum differences were determined (Tab.6, 7).All of the epochs were analysed in this way.

Final products and data integration
The following results were obtained for particular measurement epochs as a result of data processing and integration: All data and resulting products will be recorded in the designed GIS system for archaeological excavations of the Royal Castle in Warsaw.Additionally, it be amended with multitemporal images (Fig. 4A).

CONCLUSIONS
During the archaeological excavations works of the Justice Court Tower many sensors were used in different periods.
Precise TLS data (considered as reference data), as well as digital images of different resolution, acquired from different ceilings were used.All data were oriented in one reference system.The accuracy analysis of particular data, acquired by different sensors, proved that -for such analyses -it is possible to acquired metric products even when low-cost sensors are applied.The possibility to achieve such results is determined by the appropriate data processing approach.
Precision measurement is a top priority, By examining the obtained image, one may attempt to stratify chronologically the fragment of the wall.One of the methods used for this purpose is metric brick analysis.the bricks, depending on the period and the terrain, have changed, and the three-order system (length, thickness and height of the brick) allows to identify the time of brick formation.With the adoption of such a test method, the precision of measurements and the possibility of ordering the results in the form of a statistical statement is extremely important.By minimizing the error of measuring the size of bricks, we increase the probability of correctness of the obtained results.

APPENDIX
-Z+F 5006h -resolution of 6 mm/10m -performed from the inside and from the outside of the excavation -vertical images (65) with the approx.GSD value of 2 cm -the altitude of approx.54 m -automatically flight -velocity of 2.7 m/s -oblique images (144) with the approx.GSD value of 5cm -the altitude of approx.26 m -the manual flight -full-frame camera Canon 5D Mark II -no. of images -227 Epoch II -5 scans -Z+F 5006h scannerresolution of 6 mm/10m -from the inside and from the outside of the excavation No data -Olympus C-5050Z camera -no. of images -70

Figure 4 .
Figure 4. Results of integration of multitemporal data: A) the combined multitemporal image, B) the 3D model 3D and trueorthophoto from the DJI Phantom 3 Pro (Epoch I), C) the 3D model and the RGB orthoimage for the upper part of the excavation (Epoch I), D) the intensity orthoimage from TLS Z+F 5006h, and E) the 3D model from images of the central part of the excavation (Epoch II), F) the 3D model and the RGB orthoimage for the bottom part of the excavation (Epoch III)

Table 1 .
Synthesis of photogrammetry and laser scanning techniques and characteristics (Gonizzi

Table 3
scanning angles and elimination of mixed-edges effects), (d) quality control of analysed surfaces and comparison from particular scanner positions -selection of fragments in order to generate the high resolution digital surface model for the object, (e) generation of intensity orthoimages.

Table 4 .
The results of data registration

Table 5 .
The mean values of deviations smaller than 2.5 

4.2. Investigations of twist of point clouds from photographs from different epochs
Due to the monumental nature of the researched object and measurements performed in different epochs of archaeological excavations, it was decided to perform more detailed analyses of investigated fragments of surfaces of objects.It should be noticed that the object foundation (detected during excavation works) was an original construction and remaining parts of the object were reconstructed.Therefore it was decided to check mutual relations of planes registered in fragments of point clouds from different epochs.Due to relatively low mutual overlap (about 10%) of point cloud and applied protections and shelters of excavations, such analysis allows to check the mutual orientation of investigated point clouds.

Table 6 .
The average angle of plane twisting with the average max distance between planes for three epochs.

Table 7 .
The example parameters used for plane parallelism analysis (Epoch I)