Masataka Imura, Yoshito Tabata, Tomohiro Kuroda,
Yoshitsugu Manabe, Osamu Oshiro and Kunihiro Chihara
Graduate School of Information Science
Nara Institute of Science and Technology
8916-5 Takayama, Ikoma,Nara 630-0101, Japan
Tel: +81-743-72-5274 Fax: +81-743-72-5279
Faculty of Systems Engineering, Wakayama University
On January 11, 2000, Kamegata-Ishi (Turtle Shape Stone) was excavated in Asuka, Nara, Japan. Ever since, this curious-shaped stone has been attracting archaeological attention as the key to elucidate the mystery of the culture of the Asuka era. In this paper, the authors report the procedure of digital archive utilizing heterogeneous measurement methods, which are a total station, a non-contact 3D digitizer and a stereo camera. The reconstructed model holds both a detailed 3D shape and surface texture images of the stone.
What is kamegata-Ishi ?
Kamegata-Ishi (Turtle Shape Stone) is the curious-shaped stone monument, which was excavated from the Sakafune-Ishi ruins (Asuka, Nara, Japan) on January 11, 2000 (Aizawa, 2000). Figure 1 shows the appearance of the stone.
Kamegata-Ishi is located in a square that is paved with sandstone and surrounded by steps. Peculiarity of the sandstone reveals that these relics were constructed under the orders of the Empress Saimei (reigned from A.D. 655 to 661) in the Asuka era (A.D. 592-710).
Figure 1: Appearance of Kamegata-Ishi (a) Overall (b) Close-up of head
Figure 2: Procedure of restoration
The bowl-like shape of the stone indicates that the ancients in the Asuka era seemed to fill Kamegata-Ishi with water during their rite. In the Asuka era, a large number of peculiar stone monuments that seemed to be related to water were constructed. The purpose of these stones is still unknown and under discussion. The discovery of Kamegata-Ishi introduced a new clue into this remaining mystery. Therefore, the stone has been attracting a great deal of archaeological attention as the key to elucidate the culture of the Asuka era.
Procedure of Restoration
A procedure of restoration is shown in Figure 2. The procedure is mainly divided into three parts: (1) measurement, (2) data processing for each result and (3) integration.
Heterogeneous Measurement Methods
At present, a variety of tools and techniques exist for gathering 3D data of archaeological relics (Addison, 2000 and Barceló, 2000) and a combination of methods is important to obtain a precise, detailed 3D model (Nagano, 1998). The authors integrate heterogeneous measurement methods, which are a total station SET6E (Sokkia), a non-contact 3D digitizer VIVID 700 (Minolta) and a stereo camera with 35mm film.
A total station, which measures a distance to a target prism by a time-of-flight method, can cover an entire area where Kamegata-Ishi stands. However, speed of the measurement is so slow that recorded points are distributed sparsely. Moreover, no texture image can be obtained.
Therefore, to obtain detailed local shapes and texture images, the authors employ a non-contact 3D digitizer and a stereo camera. A non-contact 3D digitizer obtains local shapes, which are represented as a group of dense polygons and a texture image of the surface. A stereo camera also obtains local shapes. Although the shapes are less accurate, these texture images are more vivid than ones from non-contact 3D digitizing.
A measurement with a single device to obtain local shapes and texture images may produce an inaccurate reconstructed model because of accumulation of errors. The proposed method utilizes the data obtained by the total station as a skeleton. Consequently, accumulation of errors can be avoided in advance.
A measurement of Kamegata-Ishi was executed on March 27, 2000.
Putting Markers: First of all, markers were put on Kamegata-Ishi. Markers were used for mainly two purposes:
- Matching between a left image and a right image in stereo imaging
- Matching between a skeleton of an entire shape and local shapes for integration
Acquisition of Color and Shape: After putting markers, three measurement methods were executed simultaneously.
Figure 3: Obtained shapes (a) From optical surveying (b) From stereo imaging
The total station measured 135 points and the result was stored in a note PC connected to the total station via RS-232C. The non-contact 3D digitizer obtained 41 local shapes. Each shape consists of 200 by 200 points and 400 by 400 color texture image. The stereo camera obtained 31 pairs of stereo photographs. Figure 3 shows the obtained shapes from each measurement.
To obtain a detailed shape of the entire stone, local shapes have to be placed on precise position in a skeleton of an entire shape. This registration is achieved through the process that matches each marker of the local shape to the corresponding marker of the skeleton.
Once the correspondence between markers is given, a transformation matrix, which represents the transformation from local coordinates to global coordinates, is calculated automatically by simulated annealing. The simulated annealing can decrease the sum of differences between the marker of the skeleton and the transformed marker of the local shape into global minimum.
A rendered image of the reconstructed model of Kamegata-Ishi is shown in Figure 4(a). As the output model can be also represented in VRML2.0 format, the 3D image of the stone can be seen on the Internet .
On the other hand, the information of surrounded environment is one of great cues for investigating the purpose of the stone. The authors also obtained the image of surrounded scenery from an omni directional camera and projected the reconstructed model with scenery onto a cylindrical immersive screen (diameter 6.0m, height 2.7m) at Nara Institute of Science and Technology. The projected image can bring the higher presence to observers (Figure 4(b)).
Figure 4: (a) Reconstructed model of Kamegata-Ishi (b) Projection on immersive display
The heterogeneous measurement methods can measure a shape and a texture image of Kamegata-Ishi in a limited time and reconstruct a detailed 3D model by semi-automatic integration process.
However, some problems still remain. First, search of markers is not fully automated. To make the search automatic, the authors consider that positions of measurement must be known. The knowledge of positions limits a range of searching for corresponding markers and the variation of color of markers enables identification of one-to-one correspondence. Secondly, change of lighting condition influences color tone of texture images. To solve this patching problem, the lighting condition must be made uniform during measurement, or color of a texture image must be adjusted gradually at the edge of the texture image in the way to match with the texture image of neighboring local shapes.
The authors archived Kamegata-Ishi (Turtle Shape Stone) digitally utilizing the data fusion from heterogeneous measurement methods, which are a total station, a non-contact 3D digitizer and a stereo camera.
The heterogeneous measurement methods made it possible to reconstruct a detailed 3D shape and surface texture images of the entire stone.
The semi-automatic registration between local models and a skeleton of an entire shape reduced sharply an amount of trials and errors in integration.
This research is in cooperation with Asuka Village (Nara, Japan) and Nara National Cultural Properties Research Institute.
- Addison, A. C. and Gaiani, M., 2000. Virtualized architectural heritage, IEEE MultiMedia, 7(2), pp. 26-31.
- Aihara, Y., 2000. Turtle shape stone and Sakafune-Ishi ruins, Proceeding of Symposium “Ancient Gardens Explored Presently”, pp. 20-29 [in Japanese].
- Barceló, J. A., 2000. Visualizing what might be: an introduction to virtual reality techniques in archaeology. In: Virtual Reality in Archaeology, edited by Barceló, J. A., et al., Archaeopress, Oxford, pp. 9-35.
- Nagano, M. et al., 1998. Shape integration for 3D reconstruction of archaeological sites, International Archives of Photogrammetry and Remote Sensing, 32(5), pp. 200-207.