hul 2004 
ATROSU 
E PROJESI 
3l 
A DALI 
ATROSU 
VE PROJESI 
Es 
IM DALI 
/foto 
2002 
2002 
kim 2002 
OBJECT TRACKING AND POSITIONING ON VIDEO IMAGES 
Chi-Farn Chen, Min-Hsin Chen 
Center for Space and Remote Sensing Research, National Central University, Chung Li, TAIWAN 
cfchen@csrsr.ncu.edu.tw 
— 
Commission PS ICWG V/III 
KEY WORDS: Video, Target, Tracking, Object, Matching 
ABSTRACT: 
This study presents an application of digital video camera for object tracking and positioning. The main purpose of the study is to 
automatically trace and position a motionless object on the video images that are recorded on a moving vehicle. Because of the 
change of the focal length of camera and the movement of the vehicle, the appearance of a motionless object on the video images 
firstly will gradually change its shape and completely lose its trace after a period of time. The variation of the shape apparently will 
bring about the complexity for developing the automated algorithm to trace and position object on the video image. In this study, we 
develop a shape-based tracking technique to implement the tracking task. The technique uses the shape matrix algorithm (SMA) that 
has scale and rotation invariant characteristics to calculate the similarity of variant shapes between adjacent video frames. After the 
object is traced in every video frame, the photogrammetric collinearity condition equations are used to transform the object from the 
image coordinates to the ground coordinates. An experiment is performed to trace a motionless ship in the open sea. The result 
shows that the proposed method can successfully trace and position the ship even the ship had become entirely out of shape on the 
video images. 
1. INTRODUCTION 
Recently the digital video (DV) camera has become a popular 
monitoring tool because it is smaller, lighter and easier using 
than traditional one. By combining DV with a moving vehicle 
(e.g., helicopter or airplane), it can record a sequence of images 
conveniently and effectively. This study presents a delicate 
algorithm that can automatically trace and position a motionless 
object on the video images, which are recorded on a moving 
helicopter. The proposed method may be divided into three 
sequent steps: (1) the acquisition of the position and angular 
orientation of the camera; (2) the estimation of the image 
coordinates of the object; (3) the construction of the ground 
coordinates of the object. The first step is designed to find out 
the position and angular parameters of the video camera. In 
order to meet the strict and complex recording environment on 
the helicopter, a hardware system integrates tilt-meter, GPS, and 
digital compass with DV is developed in this study. The main 
function of the system is to record the video images and the 
camera orientation synchronously. The second step is to 
estimate the image coordinates of the object on the video images. 
The main goal of the step is to implement the task of object 
tracking on the video images. Since the object in our study 
changes the shape all the way in the recording period, this study 
uses color-based segmentation (Pei, 1999) and shape-based 
matching algorithm (Flusser, 1992; Flusser, 1995) to trace and 
locate the object. The final step is to calculate the ground 
coordinates of the object on the image sequences. Linking up 
the camera's orientations (from steopl) with object image 
coordinates (from step 2), the collinearity condition equations 
can be used to calculate the ground coordinates of the object. 
The organization of this paper is as follows. Section 2 introduces 
the information of the hardware system. Section 3 describes 
the proposed object segmentation process. In section 4, the 
shape-based object matching and tracking techniques are shown 
in detail. The positioning method will be shown in section 5. 
Finally, the experiment results and conclusions are addressed in 
section 6 and section7. 
2. THE HARDWARE SYSTEM 
In order to record the video images and orientation information 
of the camera synchronously, wherefore this study integrates 
GPS, digital compass and the tilt meter with DV through an 
encoded-decoded hardware device. The type of digital video 
camera used in recording data is SONY PC115. The accuracy 
of the tilt meter is +/- 1° with in a range of +/- 20° from the 
horizon, the accuracy of the digital compass is about +/- 3°, 
and the GPS accuracy is about 10~15 meters. The integrated 
system basically transfer the orientation data from digital format, 
acquires from the devices mentioned above, into analog voice. 
Therefore the video data and orientation parameters can be 
recorded through both video and audio channels synchronously. 
Eventually, after the recording procedure is accomplished, by 
making use of the decoded device and Microsoft DirectX 
component, the camera’s orientation and the corresponding 
video signals can be reconstructed from analog data to digital 
images.