The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B4. Beijing 2008 
dissimilarities between the seamless mapping system and 
traditional mapping system. Firstly, seamless mapping system is 
a distributed system based on network technology. Secondly, 
the workflow management are added to the system. Finally, 
during the measurement, the mosaics of digital line graphs in 
the overlapped areas by different operators are done by the 
system automatically without be intervened in by operators. 
These characteristics of the seamless mapping system are help 
to improve the efficiency of mapping. 
In section 2, the architecture of seamless mapping system is 
described and the functions of the corresponding parts are also 
introduced. In section 3, the workflows of the seamless 
mapping system is discussed in detail. And the corresponding 
effective figures of different stages in the whole workflow are 
given. In the final section, we draw a conclusion that the 
efficiency of mapping from high resolution images can be 
improved by applying the seamless mapping system. 
2. ARCHITECTURE 
A Seamless Mapping System is build based on C/S 
(Client/Server) architecture. The whole system consists of a 
single server and multiple workstations. Server and 
workstations are connected by an underground high-speed or 
fast-speed network. The whole system is configured as the 
figure 1. 
workstation workstation workstation workstation 
Figure 1. Seamless mapping system architecture 
Server acts as the following roles in the whole seamless 
mapping system: automatic processing, depository of data and 
task scheduling. Sever is accessed by workstation and data 
corresponding to tasks are automatically download to 
workstations. Then these tasks are done by operators in their 
workstations and the final mapping products are submitted to 
server. With the collaboration between server and workstation, 
workstation and workstation, and workstation and operator, the 
mapping tasks are divided two parts: automatic processing and 
human-computer interactive processing. By separating the 
automatic processing and manual measurement a mapping task 
is more efficiently accomplished by comparing to processing 
workflow in traditional DPW (Digital Photogrammetry 
Workstation). The improvement of efficiency is more obvious 
in the case of mapping from high resolution satellite imagery. 
2.1 Server 
Server usually is a high performance PC equipped with large 
volume of storage. All automatic data processing tasks such as 
image pre-processing, automatic triangulation, DSM generation 
and orthorectification are processed on server. Usually these 
automatic processing tasks are parallel processed by a blade 
cluster system or a PC cluster at the server end. The parallel 
process will improve the efficiency of the seamless mapping 
system. Beside of the parallel processes, project management, 
task management, task distributing and task monitoring are very 
important jobs which are done by server. Parallel processes will 
not be discussed in the paper. The paper emphasizes the latter. 
In the latter discussion, work unit is referred to task. 
Project management includes property setting, data 
organization and personnel management. The property setting is 
used to set the project parameters such as project root path, 
sensor type and automatic processing parameters. The objective 
of data organization is to establish a project environment. With 
this step server maintains all file system metadata for images 
and corresponding auxiliary data and other information, for 
example, which raw data should be included, how many stereo 
pair and which both images to form a stereo pair, etc. Personnel 
management is used to record the personnel information and set 
their roles in the mapping procedure. 
Task management is a very important job because the whole 
distributed collaboration mapping workflows in a seamless 
mapping system are based on the job. The first step in task 
management is task planning. How is a whole project divided 
into many parts which are dealt out to operators? It is the work 
for task planning. There are two strategies to be used to plan 
task in a seamless mapping system. The first strategy is 
planning task according to map of standard format. The second 
one is planning task according to image bounds. The former is 
applied to both line array image and frame image. The latter is 
just only applied to line array image. The two strategies applied 
to satellite line array imagery are shown in the figure 2 and the 
figure 3 respectively. 
Figure 2. Diagram of task planning by map of standard format