The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
2. SYSTEM’S ELEMENTS 
The system consists of two parts. Each part works individually 
but an interface connects them. The first part consists of camera, 
image processing and analysis, section of data evaluation and 
providing strategy, and dispatch centre. Camera has stationary 
position and is set up without geometric restraint (Figure 2). 
Image processing and analysis is integrated due to detect and 
extract Geospatial data from images. The data evaluation and 
providing strategy section assesses received data regarding to a 
defined strategy and finalizes a strategy of processing to be 
carried out by the second part of this system. The dispatch 
centre, indeed, transfers the strategy to the second part of the 
system. 
The second part, which is a dynamic platform, includes of 
vision, a computer so called Brain, and mechanical part. The 
vision is supplied for navigation, the brain is the centre of 
assessing and implementation of received information from the 
first part, and mechanical part is the motion compartment of the 
second part of the system. Next chapter will explain more about 
the system in details. 
3. METHODOLOGY 
Figure 1 is presenting a flow chart of the whole process of the 
system. 
The core of the system mostly had emphasised on simplicity, 
robustness, and precision. The majority of similar systems are 
dependent to GPS and INS. However, GPS and INS are well 
known systems for navigation, but still there are a number of 
limitations for utilising of these systems. For example, there is 
limitation to use the GPS in indoor sites or even in outdoor sites 
under some circumstances, or INS includes some errors (Chiang 
et al 2004). 
The navigation system is developed based on a geography and 
geometry presentation of a terrain. Two groups of stationary and 
non-stationary visions were utilised here. The stationary vision 
were utilised to monitoring constantly the whole of terrain in 
order to detect any unknown object. In addition, the stationary 
cameras are using to help navigation as well. As it was 
mentioned earlier, there is no requirement for camera 
calibration. Instead, A map of terrain and GIS of terrain were 
initially provided and stored in a defined memory. The aspect of 
implementation of a map and GIS of the terrain is that the 
acquired images can be replicated on the map in order to extract 
location of object according to the map. In addition, there is one 
more significant advantage of utilising a map and GIS of the 
terrain, which are distinguished this method. By utilising a map 
and GIS of terrain stationary cameras can be set up in any 
location without restraint. It means that the cameras can be set 
up in an unconventional way, like upside-down, very 
convergence, or upright position according to the terrain. 
The non-stationary cameras are mounted on the dynamic 
platform, which in this study was a RC model car. These set of 
cameras are constantly acquiring images in order to search a 
particular targets, which are supplied and set up in a certain 
locations at the terrain. The purpose of the targets is to assist the 
navigation of the second part of the system. 
All acquired images from cameras are converged in a computer 
was strengthen with pre-processing and post-processing 
softwares, which were specifically developed for this purpose. 
The softwares are required to analyse the acquired images and 
implement the following steps: 
1. To recognising and positioning any unknown object 
in the terrain. 
2. To defining the best route between the dynamic 
platform and unknown object. 
3. Dispatching the dynamic platform towards the 
unknown object. 
4. Analysing the acquired images, from camera was 
mounted on the dynamic platform, for positioning 
and navigation. 
5. And, to updating the GIS database. 
The stationary vision has been set up in different locations. 
Acquired images from the stationary vision are transfer to the 
image processing section. At the image processing section, the 
images are analysed and interpreted in order to define an object. 
The method of object tracking, which was used by Homainejad 
(2007), has been adopted and implemented in this project. Once 
an unknown object has been detected, the acquired images are 
compiled on an existing map of the terrain due to define the 
location of the object according to the map. Then the extracted 
data are transferred to the next section for analysing and 
providing a strategy. The adopted strategy will be transferred to 
second part of the system. In this part, once the computer 
component or Brain receives the information from the first part 
of the system, will send a proper order to the mechanical part or 
the dynamic platform. The dynamic platform, which is the 
motion part of the second part of the system, moves towards the 
location according to the strategy. While the second part is 
moving, the vision component will acquire images in order to 
search supplied targes on the terrain. The targes are supplied on 
the terrain in order to facilitate the navigation processing. Once 
a targe has been recognised, the image-processing component 
will define the distance of the second part to the targe. The next 
motion of the second part is based on the distance of the second 
part to the target. If the second part is in a defined a distance 
from the target, the second part will be pursue according to the 
defined strategy. 
4. ASSESSMENT OF THE SYSTEM AND CONCLUSION 
Two groups of tests have been conducted. At the first group, all 
tests have been carried out with a stationary camera. The 
camera has been set up at several locations. Each time, the 
camera has had different situation, but camera has to be able to 
monitor whole part of the terrain. Once camera has had 
convergence position with the terrain, next time camera has had 
an upright position with the terrain, or camera has been set up 
in upside-down position. In each situation a number of tests has 
been conducted. In each test, the system has to recognise an 
unknown object at the terrain and dispatches the dynamic 
platform to the object. Each time, the system successfully 
recognised the unknown object at the correct location and 
dispatches the dynamic platform to the object and there was no 
a single failing. This group of tests proves the applicability and 
ability of the system while employed a GIS and a map of the 
terrain. 
The second group of tests has been fulfilled in order to assess 
the ability of the system with stationary and non-stationary 
cameras. Similar the first group tests, the stationary vision 
recognised the unknown object and locates it. Then a strategy 
has been adopted. The strategy has been transferred to the 
second part of the system. Once the brain received the strategy.