613 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008 
l Vidro Image Data 
v 
Motion 
Motion 
Object 
('ompcncaüon | 
Detection 
Tracking 
Collaboration Control 
Figure 7. Moving target detection and tracking framework 
Figure 8 illustrates the main functional modules of the system. 
Motion compensation has image mosaicking and image 
registration two parallel sub-modules. Image mosaicking that 
could combine with other data mosaics the image sequence, and 
image registration calculates registration parameters or optical 
flow vectors. Motion detection includes background subtraction 
and target detection two serial sub-modules. Background 
subtraction restrains the movement of background using the 
parameters or the vectors, and target detection extracts target 
from the compensated background. Object tracking contains 
two serial sub-modules that are object modeling and object 
tracking. Object modelling constructs the model of object with 
its features. Object tracking realizes the successive locating of 
the object by utilizing methods corresponding to the model of it. 
Molina C ompraMtton MMidf 
Ob>«1 IiickJBK'M^iik 
Figure 8. Main functional modules of the system 
The advantages of this framework listed as below: 
(1) Parallel processing reduces the computation to meet the 
requirement of real-time application. 
(2) Transferring kinds of parameters instead of real data to 
minimizes the transmission bandwidth. 
(3) Users and exterior systems can conduct and monitor the 
modules through the interfaces offered by control module to 
evaluate the methods or make improvement. 
Moving target detection and tracking is a developing technique, 
and many technical methods will be invented and introduced 
for it in future. Though the methods may be diverse in forms 
and based theories, they have an identical purpose and conform 
to a regular system framework. Besides integrating the existing 
On the basis of analyzing the functional parts that motion 
compensation, motion detection and object tracking and the 
corresponding technical methods of moving target detection and 
tracking, we presented a new framework for the technique. We 
recognize that although there are connections between different 
sections of the technology, a serial processing of them is 
dispensable. We realized a parallel computation of the three 
parts by adding control and capture modules. The design of the 
framework facilitates the spatial separation of system and 
reduces the data stream transferred between different modules. 
This is meaningful to UAV application. Because a typical UAV 
system composes of aircraft and ground control station, and the 
data transferring depends on wireless communication. 
Our further work includes: 
(1) According to the framework, construct the testbed system 
to test the performance of technical methods and set the 
standard for evaluation. 
(2) Embedding the functional modules into the UAV system 
and improving them to meet the practical requirements. . 
REFERENCES 
Harris, C., Stephens, M., 1988. A Combined Corner and Edge 
Detector. Fourth A Ivey Vision Conference, Manchester UK, 
pp. 147-151 
Fischler, M.A., Bolles, R.C., 1981. Random Sample Consensus: 
A Paradigm for Model Fitting with Applications to Image 
Analysis and Automated Cartography. Communications of the 
ACM, 24(6), pp.381-395 
Weiss, Y., Adelson, E.H., 1996. A unified mixture framework 
for motion segmentation: incorporating spatial coherence and 
estimating the number of model. In Proc. IEEE Conf. on CVPR, 
pp. 321-326 
Alan, B., 2005. Handbook of Image and Video Processing. 
Elsevier, pp. 474-485 
Shi, J., Malik, J., 1998. Motion Segmentation and Tracking 
Using Normalized Cuts. Proc. Int’l Conf. Computer Vision, pp. 
1154-1160 
Huttenlocher, D.P., Noh, J.J., Rucklidge, W.J., 1993. 
Comparing Images Using the Hausdorff Distance. IEEE 
transactions on Pattern Analysis and Machine Intelligence, 
15(9), pp. 850-863 
Forsyth, D.A., Ponce, J., 2003. Computer Vision: A Modern 
Approach. Prentice Hall, pp. 380-396 
Comaniciu, D., Ramesh, V., Meer, P., 2003. Kernel-Based 
Object Tracking. IEEE transactions on Pattern Analysis and 
Machine Intelligence, 25(5): 564-577