- 76 - 
VOLUME BASED RECONSTRUCTION OF ARCHAEOLOGICAL ARTIFACTS 
Martin KAMPEL, Robert SABLATNIG, Srdan TOSOVIC 
Vienna University of Technology, 
Institute of Computer Aided Automation, Pattern Recognition and Image Processing Group 
Favoritenstr. 9, 183-2, A-1040, Vienna, Austria , 
{kampel,sab}@prip.tuwien.ac.at 
KEY WORDS: 3D, scanning, close range, archaeology, surface models. 
ABSTRACT: 
An algorithm for the automatic construction of a 3d model of archaeological vessels using two different 3d algorithms is presented. 
In archeology the determination of the exact volume of arbitrary vessels is of importance since this provides information about the 
manufacturer and the usage of the vessel. To acquire the 3d shape of objects with handles is complicated, since occlusions of the 
object's surface are introduced by the handle and can only be resolved by taking multiple views. Therefore, the 3d reconstruction is 
based on a sequence of images of the object taken from different viewpoints with different algorithms; shape from silhouette and 
shape from structered light. The output of both algorithms are then used to construct a single 3d model. 
Images for both algorithms are acquired by rotating the object on a turntable in front of a stationary camera. Then an octree 
representation of the object is built incrementally, by performing limited processing of all input images for each level of the octree. 
Beginning from the root node at the level 0 a rough model of the object is obtained quickly and is refined as the processed level of 
the octree increases. Results of the algorithm developed are presented for both synthetic and real input images. 
1. INTRODUCTION 
The combination of the Shape from Silhouette method with the 
Shape from Structutred Light method presented in this paper 
was performed within the Computer Aided Classification of 
Ceramics [11,6] project, which aims to provide an objective and 
automated method for classification and reconstruction of 
archaeological pottery. The final goal is to provide a tool which 
helps archaeologists in their classification process. 
Pottery was made in a very wide range of forms and shapes. 
The purpose of classification is to get a systematic view of the 
material found, to recognize types, and to add labels for 
additional information as a measure of quantity [15]. In this 
context, decoration of pottery is of great interest. Decoration is 
difficult to illustrate since it is a perspective projection of an 
originally spherical surface. This fact induces distortions that 
can be minimized by 'unwrapping' the surface. In order to be 
able to unwrap the surface it is necessary to have a 3d 
representation of the original surface. Furthermore, the exact 
volume of the vessel is a great interest to archaeologists too, 
since the volume estimation allows also a more precise 
classification [15]. Since pottery is manufactured on a turntable 
we use a turntable based method for the 3d-reconstruction of 
the original. To acquire images from multiple views we put the 
archaeological vessel on a turntable which rotates in front of a 
stationary camera. 
Shape from Silhouette is a method of automatic construction of 
a 3D model of an object based on a sequence of images of the 
object taken from multiple views, in which the object's silhou 
ette represents the only interesting feature of the image [19,16]. 
The object’s silhouette in each input image corresponds to a 
conic volume in the object real-world space. A 3D model of the 
object can be built by intersecting the conic volumes from all 
views, which is also called Space Carving [8]. 
Shape from Silhouette is a computationally simple algorithm (it 
employs only basic matrix operations for all transformations). It 
can be applied on objects of arbitrary shapes, including objects 
with certain concavities (like a handle of a cup), as long as the 
concavities are visible from at least one input view. This con 
dition is very hard to hold since most of the archaeological 
vessels do have concavities (like a cup for instance) that have to 
be modeled. Therefore a second, active shape determination 
method has to be used to discover all concavities. The acqui 
sition method used for estimating the 3d-shape of objects is 
shape from structured light, based on active triangulation [2]. 
Both methods requires only a camera and illumination devices 
as equipment, so they can be used to obtain a quick initial 
model of an object which can then be refined stepwise. 
There have been many works on construction of 3D models of 
objects from multiple views. Baker [1] used silhouettes of an 
object rotating on a turntable to construct a wire-frame model of 
the object. Martin and Aggarwal [10] constructed volume seg 
ment models from orthographic projection of silhouettes. Chien 
and Aggarwal [4] constructed an object's octree model from its 
three orthographic projections. Veenstra and Ahuja [22] extend 
ed this approach to thirteen standard orthographic views. 
Potmesil [16] created octree models using arbitrary views and 
perspective projection. For each of the views he constructs an 
octree representing the corresponding conic volume and then 
intersects all octrees. In contrast to this, Szeliski [19] first 
creates a low resolution octree model quickly and then refines 
this model iteratively, by intersecting each new silhouette with 
the already existing model. The last two approaches project an 
octree node into the image plane to perform the intersection 
between the octree node and the object's silhouette. Srivastava 
and Ahuja [18] in contrast, perform the intersections in 30- 
space. The work of Szeliski [19] and Niem [13] were used as a 
basis for the shape from silhouette approach presented in this 
paper. For the active triangulation method we use an approach 
bis Liska developped for a next view planing strategy using 
structured light [9]. 
The paper is organized as follows. Section 2 describes the 
equipment used. Section 3 describes the strategy for and