基于 Matlab/Simulink 脸部疲劳状态检测系统的研究
ii
A STUDY FOR THE DETECTION SYSTEM OF FACIAL
FATIGUE BASED ON MATLAB/SIMULINK
ABSTRACT
In modern times, it is necessary and meaningful to study the detection of driver’s fatigue
state and develop the aided equipments for the drivers because fatigue driving is becoming one
of the most severe factors to cause the traffic accidents. In recent years, the detecting systems
have been achieved on DSP. Although there are well known technology for the hardware, knots
are to develop the available software system, some of software are very difficult to debug and
code. In order to save the time for the source codes by R&D personnel, a model developed by
myself is presented. Based on Simulink, these models can obtain C codes easily which can run
by DSP.
Firstly, the research background of fatigue detection system and the present study of
fatigue detection in abroad and in local are presented in this thesis. And then, the developed
system is depicted in detail.
The work is described into two parts here. In the first section, at first, some necessary
preprocessing is done before face detection, such as illumination intensity compensation. Using
skin color models on the original image for detecting face, then the binary image is obtained.
After that, do some morphological treatments to eliminate noise, the facial center is located
roughly and the region is searched by using the quick projection method. According to the
facial constraints, the specific location of a human face can be located accurately. The
projection method is used to locate eyes on the upper half part of the facial region. For the
lower half part of the facial region, lips are located successfully by searching the largest
connected region method. Finally, by using the ratios of eyes and lips regions as the criteria, the
state of fatigue is detected quickly and accurately. All of them are implemented with the
MATLAB environment.
In the second section of the thesis, based on the above ideas in the first part of this work,
the detection system of facial fatigue is completed on the Simulink processing platform, some
models are carried out. And through the Build option that is in the Simulink's own Real-Time
Workshop page, a large amount of C codes available on DSP are obtained.To test our simulation
work,the debugged work is carried out on DM642 and good results are obtained.
