«LASER SCANNING ISSUES FOR THE GEOMETRICAL RECORDING OF A COMPLEX STATUE Maria TSAKIRI1, Charalambos IOANNIDIS1, Alistair CARTY2 1 School of Rural and ...»
Tsakiri, Ioannidis, Carty 1
LASER SCANNING ISSUES FOR THE GEOMETRICAL
RECORDING OF A COMPLEX STATUE
Maria TSAKIRI1, Charalambos IOANNIDIS1, Alistair CARTY2
School of Rural and Surveying Engineering, National Technical University of Athens, Greece.
Archaeoptics Ltd, Glasgow, UK
KEY WORDS: laser scanning, heritage applications, three-dimensional, close range, data capture
ABSTRACTRecent advances in laser scanning technology allow for fast and efficient 3D documentation of cultural heritage artefacts and statues. The advantages are more evident when the objects are of large size and comprise complex surfaces and fine details. There is therefore, a need to addressing clearly all the practical difficulties and issues that one has to consider when using this technology for the documentation of large archaeological statues. This paper describes a number of specific issues, from data capturing procedures to data management and processing, raised in a project aimed in producing a geometrically accurate 3D model of the ancient statue of Hermes by Praxiteles. Finally, results from the 3D solid model of the statue are also provided.
The documentation of archaeological artefacts located in non-laboratory environment such as museums, presents further challenges, mainly because the statues are permanently housed and the conditions during the data capture procedure cannot be easily altered, for example the lighting of room or the limited space around the statue. For statues of long height, data capture for the top parts of the object can be problematic for practical reasons, which include restricted working space and use of mobile workbenches. An overriding concern during data capture of large fragile objects is the safety of the statues. The techniques that can be employed should involve non-contact methods (ie avoidance of setting targets on the surface of the statue) and measuring systems that can control the levels of light andheat as these can potentially damage the artwork.
Further to data capture, the processing of the data in order to create an accurate 3D solid model of the captured object for purposes such as digital documentation, moulding or even rapid prototyping, can be an extremely difficult task because of the management of large volumes of data.
Recent advances in laser scanning technology allow for fast and efficient 3D documentation of cultural heritage artefacts and statues. A great advantage of the technique relies in that is entirely a non-contact method thus allowing for the fine geometric details of sculptured and irregular surfaces to be more easily captured when compared to close-range photogrammetric methods. Clearly, the combined use of both techniques offers many advantages (eg Bitelli et al. 2002; Henz 2002), Cultural Heritage 2 however in many applications, such as in metric cultural heritage documentation of sensitive complex statues, laser scanning is preferred to record data economically and with high accuracy.
There is therefore a need to addressing all the practical difficulties and issues about the data collection and processing that one has to consider when using this technology for the documentation of large archaeological statues.
Since a significant proportion in the world’s museums heritage collections are large complex statues, this paper aims to discuss the issues from data capturing procedures to data management and processing for the creation of the final model of such an ancient complex and fragile object using a high resolution terrestrial laser scanner. A description on the various stages of the data collection and processing to create an accurate 3D solid model of the statue of Hermes by Praxiteles is given and results of the digital 3D model of the statue are also considered.
2 DATA COLLECTION The aim of the project was to acquire a geometrically accurate 3D model of the Hermes statue, which is housed at the archaeological museum in Ancient Olympia in Greece. The area is subject to high seismic activity and the archaeological artefacts of the museum are therefore prone to vibrations generated by the earthquakes. To minimise the ground accelerations and improve the likelihood of reducing damage of sensitive structures, such as the statue of Hermes, it was decided by the museum authority to construct a seismic isolation retrofitting assembly. The assembly is designed to enhance the statue’s safety without impairing its appearance and this process required an accurate 3D model of the statue.
The Hermes statue, dated to 343 BC, is made from Parian marble and is considered a masterpiece of ancient Greek art. It is the only original work of Praxiteles that has survived and was found at Olympia, intact on his base, several metres under the ground. It was dedicated to the sacred Altis from the Eleians and Arcadians to commemorate their peace treaty. Later it was moved to the temple of Hera, where it was found in 1877 AD. The statue shows Hermes (Fig. 4) holding in his left arm the infant Dionysos while in his raised right hand he probably held a bunch of grapes, dangling it before the childgod, who stretches out his small hand to grasp the sacred fruit. The statue of 2.13m height is standing on a marble base of dimensions 1.25m in height and 1.26m by
The 3D laser scanner implemented for the data capture is a Minolta VI-900 (http://www.minoltad.com), which is a state-of-the-art system, and is typically tripod-mounted for maximum ease of use. The highest resolution the scanner can provide is 0.17mm (170 microns). It is a laser lightstripe triangulation rangefinder of short range, with typical distances from the object being in the order of 0.6-1.2m. It also provides colour data and the quoted precision is ± 0.008mm for the fine operating mode.
In the ensuing discussion, the practical aspects need to be considered when data collection is performed from a large fragile statue using a short-range laser scanner instrument are addressed.
2.1 Mounting of scanner Statues have surfaces that point in all directions with many horizontal crevices, such as the folds in carved drapery. Therefore, the scanner needs to have flexibility and be able to scan surfaces of any orientation. Also, the top parts of the statue, such as the head and upper body parts require a configuration whereby the scanner is mounted on a level above the tallest part of the statue to facilitate data capture from all directions. Scaffolding around the statue (Fig. 1) with appropriate Tsakiri, Ioannidis, Carty 3 levels to allow scanning of the tops of the statue and the difficult regions, which are hidden or occluded by parts of the statue, was constructed. It is important to note the stability of the scaffolding during data capture so that vibrations of the system, mainly caused by the operator, are minimised.
Fig. 1. Scaffolding levels to secure access and stability of laser scanner An overriding concern during data capture is to avoid harming the statue by accidental collisions between the scanner and the statue, as well as being tangled by the instrumentation cabling. The wide scaffolding, which was used in this project and consisted of wooden floors with a width of 1m around the statue and at a distance of 0.2m from the outline of its basis, prevented such problems.
When the levels of scaffolding are not wide enough, accidents may be caused by tripod leg deployment during scanner repositioning. It is therefore necessary to have a “scan plan” that minimises the number of scanner positions and thereby reduces the potential for colliding with the statue.
The Minolta scanner's motion platform is a customised heavy-duty camera tripod head that allows the scanner to be reoriented through pan, tilt and translate motions enabling difficult to reach areas to be scanned. This also minimises the number of repositions of the scanner platform as many scans can be taken from the same physical position by simply reorienting the scanner on the tripod head.
2.2 Scanning procedure
Using the Minolta VI-900 scanner, the scanning procedure requires an operator who interactively moves the scan head to set each new scanning window. The window is constrained by both the field of view of the lens being currently used and any occlusions of either the laser or the line-of-sight camera both mounted on the front of the scanner. After setting up the scan, the actual process takes three seconds to acquire roughly 300,000 points. RGB colour data at a per-vertex level is acquired immediately after the scan is complete via the camera used for triangulation. The RGB data for the Hermes statue was not required to be photo-realistic, merely diagnostic for determining material changes and breaklines in the statue.
However, any tremors either caused by the pressing of the scanner “Release” button or through movement of the scaffolding may introduce folds in the range data. Such an example is shown in Fig. 2, which depicts “rippling” on the statue’s chest due to tremors in the scanner platform.
Cultural Heritage 4 Fig. 2. “Rippling” on the chest due to tremors in the scanner platform All scans for the Hermes statue were obtained at a distance of about 1m from the statue and at
0.5mm resolution with an accuracy of 0.25mm. This resolution was decided upon in order to achieve the goal of recording the intricate details like chisel marks smaller than a millimetre and any changes of material but without the dataset becoming completely unmanageable in size.
Clearly, for capturing complex surfaces, a large number of scan images are needed to cover the object. The number of scans required depends on the shape of object, amount of self-occlusion, the object size and the desired resolution of the source dataset.
The duration of the Hermes scanning was two days, during which multiple overlapping scans were obtained. The overlap of adjacent scans was about 20%-30%, which not only enables registration to work more efficiently but also helps quantify any distortive effects on accuracy the underlying matrix was causing.
Regarding safety of the statue from the scanning process as a non-contact method, laser triangulation does not cause harm because only light touches the object. The scanning beam was a 690nm red semiconductor laser and it moved continuously during scanning. The difficulty of scanning under field (non-laboratory) conditions cannot be underestimated considering also the fact that the data capture was performed during restricted time periods following the museum’s operating hours.
2.3 Marble surface
The type of material of the object to be scanned may significantly affect the captured laser data. The underlying hypothesis of active optical geometric measurements is that the imaged surface is opaque and diffusely reflecting. The marble, which is the material used for the Hermes statue, is composed of densely packed transparent crystals causing it to exhibit subsurface scattering. Marble departs from this hypothesis and exhibits two important properties of translucency and nonhomogeneity (Godin et al. 2001). These properties generate two key effects on the geometric measurement and result in a bias in the distance measurement and an increase in noise level when compared to a reference opaque surface.
Tsakiri, Ioannidis, Carty 5 Fig. 3. The right knee. Beneath the knee is Plaster, above is Parian Marble. The contrast in noise levels is quite marked Specifically, the shiny marble, by causing some increase in the subsurface scattering, may result in degradation of the quality of the range data and disturb the way that the models are rendered.
Levoy et al. (2000) state that their assumption on treating the diffuse reflectance as ideal Lambertian when scanning marble statues was possibly not correct because the computed reflectances from each scanned surface did not agree since the laser observations included noise and other errors. In scanning the Hermes marble statue, it was noticed that the reflected spot was shifted away from the laser source and also the spot varied in shape across the surfaces of the statue due to the random crystalline structure of the marble leading to noise in the depth values. In fact, the surface made of plaster, which is located at the right knee of Hermes, did not exhibit the same reflectance effect compared to the polished marble surfaces, as seen in Fig. 3. For this reason, a number of data test collections from the statue were performed in order to assess the material properties and eliminate the noise in the final model.
2.4 Data management The datasets produced by short range, high resolution laser scanners during the scanning process can quickly become impossible to handle and unmanageable for post-processing. There are many different techniques on storing datasets, with most commonly being the storage as point clouds, polygon meshes or range images. Individual meshing or automatic triangulation are often applied on the whole point cloud to create models but extremely high performance from a PC is required to achieve acceptable speed when executing such calculations. On the other hand, range images are more efficient and commonly used in computer graphics because the scanned surfaces are stored as encoded range images (Pulli 1999).