Dynamic Video Tracking Monitoring

System Based on Human-computer Interaction

Technology

Wei Song1 Ming Li2

College of Physics and Electronic Information College of Physics and Electronic Information
Tianjin Normal University Tianjin Normal University
Tianjin, China Tianjin, China
[email protected] [email protected]

Abstract—This design is purposed to develop a kind of makes the system more convenient, flexible and interface
dynamic video tracking monitoring system involved video friendly.
tracking technology, 3G wireless transmission network, and Besides, the key technology of this system is to
intelligent human-computer interaction based on touch improve detection efficiency and accuracy by using the
screen technology. This design has introduced the intelligent target detection algorithm based on time and space
human-computer interaction based on touch screen dependence. Traditional target detection algorithms mainly
technology, which benefits users to lock the target and include interframe difference algorithm, background
relocate it when it gets lost. The hardware configurations of difference algorithm and optical flow algorithm. Those
this system include monitoring-side and server-side. To
three algorithms have not simultaneously considered both
increase the detection efficiency and accuracy of this system,
space and time characteristic of an image, which leads an
this design creatively puts forward the target detection
algorithm based on time and space dependence to conduct
undesirable and nonintegrated detecting to slight
video tracking at the monitoring-side. Until now, it is rare to movement of a target moving in complex background. To
see such products or designs that use similar technologies. settle the deficiency above, this design has proposed a kind
of modified algorithm based on time and space
Keywords-human-computer interaction; dynamic video; dependence, which increases the system detection
touch screen; 3G; the target detection algorithm based on time efficiency and accuracy to great extent.
and space dependence Moreover, this design uses 3G (or high speed wireless
transmission network) to conduct long-distance video
I. INTRODUCTION monitoring, which improves the real time of monitoring
target of this system greatly.
Most of current video monitoring systems usually use
traditional video monitoring equipment, whose II. EXISTING PROBLEM
constitution is complex, and operating is onerous; It is a tendency to develop dynamic video tracking
moreover, they generally use wired method to transmit monitoring system with high detection efficiency and
video, which results that the monitoring-side is not easy to accuracy. Recently, mainly three algorithms have been
move flexibly. utilized in most monitoring systems to improve their
Dynamic video tracking monitoring system is the detection efficiency. They are the detection algorithm
tendency of future development of video monitoring based on the dependence of continuous frames of moving
system, and it will be widely used in areas including target, the detection algorithm based on the dependence of
traffic, security, education, fire control, medicine, and so changing background of moving target, or optical flow
on. Nowadays, current security industry has devoted too algorithm. But in these three algorithms, it is inevitable
much sources and time to research and develop dynamic that the detecting to target moving at complex background
video transmission technology. Though partial applications or the detecting to target whose color is similar to that of
of this technology into monitoring area have been its background is undesirable and not integrated, because
developed, which could conduct video transmission by none of those three algorithms considers both the space
using wireless broadband technology or 3G networks, the and time characteristic of moving target simultaneously.
integral remote monitoring application platform has not
been developed. III. THE TARGET DETECTION ALGORITHM BASED
To settle the problems above, this design has conducted ON TIME AND SPACE DEPENDENCE
redesign based on traditional video monitoring technology.
It has added the video tracking and intelligent human- The principle of this algorithm: there exists strong time
computer function based on touch screen technology to and space dependence between continuous video image
control Pan/Tilt/Zoom (PTZ) and lock the target, which frames, and it will produce great effect if dividing video
image by taking advantage of this space-time dependence.
In one frame, there always exists space continuity among
the colors of pixels that record scenery: the values of
colors in adjacent areas change slowly; the characteristics
of brightness or texture in adjacent areas are similar. Those
are the space dependence of an image.
The time dependence reflects in the point where a
certain pixel in one frame of a video will definitely occur
in a specific location of the next frame, while its
corresponding pixel in this later frame is unknown.
The procedure of the target detection algorithm based
on time and space dependence:
• By taking advantage of the space dependence of
moving target, dividing the interframe difference
image into blocks, and conducting difference
operation by using the statistical average
brightness value of one integrated image block but
not the brightness value of a single pixel, then
judging whether this whole block is part of this
moving target or not through the threshold value
comparison. Taking the whole block but not a
single pixel as the object of difference operation
can effectively prevent certain pixels from not
being checked because their colors are similar to
that of their background.
• By taking advantage of the time dependence,
defining certain range around current target, then
affirming that there indeed exists moving target in
this range only when the same target can be
detected in this range at next time. If not, it will be Figure 1. system composition figure
defined as noise.
As a matter of fact, not any technology can be 100%
accurate. On the basis above, this system introduces the V. HARDWARE CONSTITUTION
intelligent human-computer interaction interface based on The video image collected by this system is
touch screen technology. It can relocate the target by transformed to digital signal through video A/D, and then
means of human-computer interaction if target gets lost. transmitted to image processing DSP unit through
Complex Programmable Logic Device (CPLD). The signal
IV. SYSTEM CONSTITUTION disposed by CPLD also could be transformed to VGA
This system consists of monitoring-side and server- signal through D/A, then be sent to the display supporting
side. As shown in figure below, monitoring-side conducts VGA interface. Image processing unit DSP could collect
video tracking by means of the target detection algorithm the human-computer interaction information through SPI
based on time and space dependence. Considering it has bus, analyze transmitted image information by using the
introduced the intelligent human-computer interaction target detection algorithm based on time and space
interface based on touch screen technology, this design dependence, synthesize the human-computer interaction
could conduct target location by relying on the method of information, figure out the characteristics of target, and
human-computer interaction when the target gets lost. The finally send the tracking instruction to control high speed
video data of monitoring-side could be transmitted to PTZ to conduct accurate tracking. Moreover, image
server-side through 3G (or high speed wireless processing unit DSP could store part of image information
transmission network), and then to be observed by server- into its FLASH. Here is the hardware structure of this
side real-timely. system.
 Figure 3. program flow chart

VII. RESULTS
Figure 2. hardware structure diagram On the basis of these hardware components and
software structure, the improved tracking detection
algorithm increases the detection efficiency and accuracy
VI. SOFTWARE PROCESS of this system and guarantees the integrity of tracking the
After power on reset, the system checks all its parts target greatly.
according to protocol automatically to guarantee the Our design takes the target detection algorithm based
normal running and communication of each part. The on time and space dependence as the method to track
default state of PTZ is silence state, which means that PTZ target. This algorithm overcomes the deficiency of
will observe the scope focused by camera after the self- commonly used algorithms that the detecting to slight
checking. Besides, the server-side could also receive image movement of target moving in complex background is
in the same fixed range. undesirable and not integrated. It improves the detection
The system waits users to choose target after the efficiency of this system to great extent.
completion of self-checking. Once the target is confirmed, Moreover, in order to settle the problem of losing
the DSP unit will conduct high-speed operation to target, our design has introduced the intelligent human-
calculate PTZ control instructions by using the target computer interaction interface based on touch screen
detection algorithm based on time and space dependence. technology, which benefits users to relocate the target
That is to control PTZ to film the target real-timely and when it gets lost by using the method the human-computer
accurately. interaction.
Considering the limitation of this system and the
complexity of tracking background, the target may get lost. VIII. THE SCOPE OF APPLICATION
So this system has designed a certain function to deal with The scope of application of this design is very wide,
lost target. When the system loses its target, it will alert to including traffic, public security, education, fire control,
indicate users that the target gets lost, so users could medicine, and so on. Here is its concrete application in
relocate the target immediately by using the human- tracking target:
computer interaction interface. Moreover, the lost Firstly, this system could track target selected by users.
information could also be transmitted back to the server- Users could use the human-computer interaction interface
side. Here is the program flow of this system.
to locate the target. Then the PTZ will lock this target to system [J]. Jornal of Shanghai Jiao Tong Univrsity, 2006, 36 (12):
conduct tracking filming. 1837-1840.
Secondly, this system could relocate the target when it [3] Yongxin Liu, Ping Wei, Chaozhun Hou. The rapid method of
moving target detection in video image [J]. Jornal of
gets lost. When the target gets lost, users could also use the Instrumentation, 2002,23 (5): 163-166, 176.
human-computer interaction interface to find it and [4] Meier T, Nguo K N. Video Segmentation for Content-Based
relocate it. Coding [J].IEEE Trans on Circuits and Systems for Video
technology. 1999,9(8):1190-1203.
[5] Lipton A Fujiyoshi H, Patil R. Moving target classification and
REFERENCES tracking from real-time video [J]. Proc of WACV’98,1998: 8-14.
[1] Stauffer C, Grimson W. Adaptive background mixture models for [6] Hui Zhu, ZaiMing Li. The moving target detection technology in
real-time tracking [J].Proceedings of IEEE Computer Society video order [J]. Signal Processing. 2002,18 (5): 448-451.
Conference on Computer Vision and Pattern [7] Xinquan Tan, Boyue Li, Huafei Zheng, etc.The research of real-
Recognition,1999,2(6):248-252 time monitoring target in background noise.
[2] Wenjie Zhang, Feihu Qi, Zhuojun, Jiang. One practical method
about target detection and tracking of real time video monitoring