09-08-2014, 01:00 PM
Embedded Remote Video Surveillance System Based on ARM
[attachment=66804]
Abstract
In this study, an embedded remote video surveillance system based on ARM was build. The
hardware system is composed of ARM S3C2440 processor, OV9650 CMOS camera, and mobile phone;
and the software platform includes three modules——video capture, video processing, and video
transmission. Moreover, an improved moving target detection algorithm was proposed, which combines
real-time background updating and three frame differencing to reduce the impact from the gradient and
emergency in ambient light when long-term monitoring. The test results showed that, when someone
appears in the monitoring scene, the system will send the relevant video information to the user’s mobile
phone via e-mail, that is, an alarm is sent to the user to complete remote video surveillance. In addition, it
can adapt to the changes in ambient light when long-term monitoring and also has a good resistance to
the sudden appearance of highlight, having good security and stability
INTRODUCTION
The video surveillance system based on computer
technology, embedded technology, and network technology
can carry on all-day automatic real-time monitoring to the
important department or place of the various professions, and
it has already become the important component of peaceful
guard domain. In recent years, the research on video
surveillance system has caused widespread attention of the
academic circle and industrial field in home and abroad,
which also becomes the front topic of domestic and foreign
research (Xing et al, 2006; Xiong et al, 2009).
In 1997, the first remote monitoring diagnosis workshop was
sponsored untidily by the Standford University and
Massachusetts Institute of Technology, and there were more
than 50 representors from 30 corporations and research
facilities attending the conference. The conference mainly
discussed the open style system, the diagnostic message
regulations, the transport protocols, and legitimate limit to
users, and so on, which are related to remote surveillance
system, and made forecast to the future technological
development. The next generation remote monitoring
diagnosis demonstration system based on Internet was
developed by the Standford University and Massachusetts
Institute of Technology, which has been supported and
coordinated by 12 big companies about manufacturing
industry, computer industry, instruments, and meters’
industry, such as Sun, HP, Boeing, Intel, and Ford, and so on.
Afterwards, these companies promoted an experimental
system named Testbed together. Testbed uses the embedde
SYSTEM DESIGN
The hardware system includes processor, video-capture
devices, and user’s mobile phone to receive video
information. In this study, ARM920T-based S3C2440
processor is chosen to complete the core control; OV9650
CMOS camera is used as a video-capture device; and the
user’s phone is connected to the Internet to receive video
information to achieve real-time monitoring. As shown in
Fig.1, it is the block diagram for overall design of the
hardware system
Design of Software Platform
The embedded WinCE operating system (Pang et al, 2009) is
selected in the design, and there are mainly three function
modules, that is, video capture module, video processing
module, and video transmission module. The schematic
diagram of system software platform is shown in Fig.2.
The main function of each module is as follows:
(1) Video capture module: Camera captures video frames and
sends them to video processing module.
(2) Video processing module: Use Moving target detection to
process each video frame to determine whether there is
moving target in the monitoring scene.
DESIGN OF MAIN FUNCTION MODULES
Video capture module
CMOS OV9650 camera is selected to capture video frame,
the video format is RGB565, which represents a pixel by 16-
bit (2 bytes). The lowest 5 bits denote blue component, the
middle 6 bits denote green component, and the highest 5 bits
denote red component. The masks corresponding to the three
components are: 0xF800, 0x07E0 and 0x001F. In order to
extract the desired colour component, after reading a pixel,
use the respective mask “and” on the pixel value. The flow
chart of video capture is shown in Fig.3
Video Processing Module
his module handles video information from video capture
module, that is to say, detecting whether there is an intruder
appearing in the monitoring scene.
In order to improve accuracy and stability of the system, on
the basis of background subtraction, an improved moving
detection algorithm is proposed. The main idea is: in
background subtraction, the background image is fixed, so in
the case of long-term monitoring, the changes of ambient
light will reduce the accuracy of moving detection. To
overcome this drawback, the adaptive background updating
method is introduced, which can real-time update the
background to reduce the impact of the ambient light.
However, it has poor adaptability to the sudden changes in
the environment, such as sudden exposure, which can also
result in
SYSTEM EVALUATION AND EXPERIMENTAL
Use “eMbedded Visual C++ (eVC++) 4.0” to develop the
local applications, and select “Intelligent Control and
Embedded System Laboratory, College of Computer Science
and Technology, Jilin University” to be as the monitoring
scene. On this basis, the performance of our embedded
remote video surveillance system is tested. In the test
program, set the output format to be YUV422, the rate of
capturing frame to be 25f/s, and the resolution to be 320× 240.
As shown in Fig.7, it is the interface map of our system
starting.
In the system, each video frame is saved with the fixed size,
and in the video processing module, by comparing the
number of pixels in the continuous region of the difference
image and the threshold T, it can be determined whether there
is someone braking into the monitoring scene. After several
experiments, the threshold T is chosen to be a fixed value
CONCLUSIONS
In this study, an embedded remote video surveillance system
based on ARM was built, embedded WinCE operating
system was selected; the video capture device is made up of
S3C2440 processor and CMOS camera (OV9650) with 1.3
million pixels, and the alarm message is sent to the user’s
mobile phone via e-mail. In addition, combining with
background subtraction, frame difference and real-time
background updating, an improved moving target detection
algorithm was proposed, which enhances the stability and
accuracy of our system.
Our system has the following advantages: (1) in ARM
system, use CMOS image sensor to capture images, their
color and quality are acceptable, although imaging
permeability and color reproduction are not as CCD, it has
advantages at low power consumption, price and integration;
(2) the video processing module has good real-time
processing performance and the saved video frames have
high quality; (3) taking the amount of information
transmission and different types of mobile phones, and using
mobile mailbox to send alarm message, is better in
adaptability, faster at speed and lower at price.
Of course, there are still some shortcomings in the system:
(1) the robustness has not yet reached the requirements of a
complete intelligent monitoring system, for example, when
the speed of the moving target changes sharply, there may be
error detection; (2) the system was designed based on wired
network, there may be wiring problems, if being extended to
wireless network, its application will be more convenient; (3)
using mobile mailbox to send alarm message is slightly
inferior to phones with good performance and full function,
the communication with mobile phone by GPRS will be more
direct and convenient.