rapid 
Or 
jte 
are 
The 
)), 
her 
der 
ial 
ing a 
le 
nitive 
ry of 
1989, 
hing. 
mote 
ange 
mera. 
>mote 
1997, 
multi- 
1990, 
logy, 
  
  
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 5. Hakodate 1998 
SOME TESTS OF TLS IMAGE MATCHING FOR RECONSTRUCTION OF 3D ARCHAEOLOGICAL OBJECTS 
Xiaoyong CHEN 
Senior Researcher, Asia Air Survey Co., LTD. 
13-16, Tamura-Cho, Atsugi-Shi, Kanagawa 243, JAPAN 
E-mail: xy.chen @ ajiko.co.jp 
Shunji MURAI 
Chair Professor, STAR Program, Asian Institute of Technology 
P.O. Box 4 Klong Luang Pathumthani, THAILAND 
E-mail: murai@ait.ac.th 
Commission V, Working Group V/5 
KEY WORDS: Archaeological Object, TLS Image Matching, Orthoimage Generation, 3D Object Modeling 
ABSTRACT 
Three Line Sensor (TLS) systems, as a kind of new technique for digitizing and recording of archaeological objects with 
high accuracy and resolution, have been widely used in archaeological field in recent years. For reconstruction of 3D 
spatial models, the algorithms and system of TLS image matching have been developed and reported by Wu and Murai in 
1996. In this paper, we reported some of our results for using Wu and Murai's system and discussed how to make 
improvements for our applications. Three parts of works have been reported, which include how to use the system for real 
large TLS image matching, how to fast generate orthoimages based on 3D models, and how to create 3D virtual 
archaeological objects for visualization. Finally, some examples are given to show the efficiency of our approaches. 
1. INTRODUCTION 
Recently, there are considerable interests in automated 
extraction of 3D information from archaeological objects. 
Digital close-range photogrammetry based on the high- 
resolution images captured by CCD three line scanners 
(TLS) has played very important role in the field of 
recording and visualization of cultural heritages. One of 
central tasks in digital photogrammetry and computer 
vision is to reconstruct the forms and structures of objects 
by image matching based on two or more images in the 
scene. During the recent decades, many image matching 
techniques (such as area based matching, feature based 
matching, and relation based matching) have been 
investigated by many researchers both in the fields of 
photogrammetry and computer since, but there are still 
many unsolved problems need to be improved. 
  
For the reconstruction of archaeological objects using 
linear CCD camera, the algorithms and system of image 
matching have been developed and reported by Wu and 
Murai in 1996. The image data were obtained by a triplet 
of linear CCD cameras (Three Line Scanner) and 
registered to satisfy the epipolar constraints. The 
advantages of this system are mainly included as follows: 
1) It can generate a dense disparity map for accurately 
measuring 3D objects; 
2) The algorithms are robust due to eliminate 
mismatches using the extra images; 
3) The algorithms are based on the disparity constraint 
of the TLS images and without needs explicit 
knowledge of the camera's parameters. 
In this paper, we reported some of our testing results for 
" The author is now working as a visiting faculty in STAR Program of Asian Institute of Technology, THAILAND 
521