07-12-2012, 05:54 PM
A Low-cost Extendable Framework for Embedded Smart Car
Security System
[attachment=43010]
Abstract—
In this paper, a low-cost extendable framework for
embedded smart car security system is proposed, which consists
of a face detection subsystem, a GPS (Global Positioning
System) module, a GSM (Global System for Mobile
Communications) module and a control platform. The face
detection subsystem bases on optimized AdaBoost algorithm
and can detect faces in cars during the period in which nobody
should be in the car, and make an alarm loudly or soundlessly.
The other modules transmit necessary information to users and
help to keep eyes on cars all the time, even when the car is lost.
This system prototype is built on the base of one embedded
platform in which one SoC named “SEP4020” (works at
100MHz) controls all the processes. Experimental results
illuminate the validity of this car security system, and it is also
much cheaper and ‘smarter’ than traditional ones.
INTRODUCTION
wITH the development and applications of many
embedded techniques, car security system design and
analysis are constantly improving. Many new techniques,
such as biometric recognition technique, image processing
technique, communication technique and so on, have been
integrated into car security systems [1] [2]. At the same time,
the amount of accident of cars still remains high, specially,
lost. So, one practicable car security system should be
efficient, robust and reliable.
Face detection process
Face detection is to find faces in one image by the trained
cascade classifiers. Every node determines whether there are
faces in the image according the data in classifiers’ data file
which is the outcome of training process. As a result, face
detection process is a pure calculation process, and most of
the results of face detection research papers are obtained by
detecting images on personal computer platform. But in the
low-cost extendable embedded smart car security system, no
powerful CPU could be utilized. In several papers of the
recent years, DSP (Digital Signal Processor) or FPGA (Field
Programmable Gate Array) are used to speedup the detection
process and to meet the real-time target, but the cost of whole
system is increased at the same time. In car security system,
the demand of “real-time” may not be as rigorous as other
application environment, such as to distinguish a criminal out
of people in street. Since the driver will not leave the car in a
very short time, the car security system have a few seconds to
make the judgment of the face is detected, and the period of
time is enough long for the car security system to accomplish
the face detection process.
Security System
[attachment=43010]
Abstract—
In this paper, a low-cost extendable framework for
embedded smart car security system is proposed, which consists
of a face detection subsystem, a GPS (Global Positioning
System) module, a GSM (Global System for Mobile
Communications) module and a control platform. The face
detection subsystem bases on optimized AdaBoost algorithm
and can detect faces in cars during the period in which nobody
should be in the car, and make an alarm loudly or soundlessly.
The other modules transmit necessary information to users and
help to keep eyes on cars all the time, even when the car is lost.
This system prototype is built on the base of one embedded
platform in which one SoC named “SEP4020” (works at
100MHz) controls all the processes. Experimental results
illuminate the validity of this car security system, and it is also
much cheaper and ‘smarter’ than traditional ones.
INTRODUCTION
wITH the development and applications of many
embedded techniques, car security system design and
analysis are constantly improving. Many new techniques,
such as biometric recognition technique, image processing
technique, communication technique and so on, have been
integrated into car security systems [1] [2]. At the same time,
the amount of accident of cars still remains high, specially,
lost. So, one practicable car security system should be
efficient, robust and reliable.
Face detection process
Face detection is to find faces in one image by the trained
cascade classifiers. Every node determines whether there are
faces in the image according the data in classifiers’ data file
which is the outcome of training process. As a result, face
detection process is a pure calculation process, and most of
the results of face detection research papers are obtained by
detecting images on personal computer platform. But in the
low-cost extendable embedded smart car security system, no
powerful CPU could be utilized. In several papers of the
recent years, DSP (Digital Signal Processor) or FPGA (Field
Programmable Gate Array) are used to speedup the detection
process and to meet the real-time target, but the cost of whole
system is increased at the same time. In car security system,
the demand of “real-time” may not be as rigorous as other
application environment, such as to distinguish a criminal out
of people in street. Since the driver will not leave the car in a
very short time, the car security system have a few seconds to
make the judgment of the face is detected, and the period of
time is enough long for the car security system to accomplish
the face detection process.