Professional Documents
Culture Documents
AbstractAn innovative automatic door control system is latter analysis for the applications such as crime investigation
proposed in this paper to overcome the drawback of frequent and prevention.
false actions among the present devices while increasing their
added values for security applications. Through human This paper is organized as follows: Section II describes
detection and trajectory tracking techniques, the proposed the system design concept and related theoretical basis, while
system can precisely identify those people with intention of the corresponding hardware implementation architecture is
entering or leaving the door, and then control opening and described in section III. Section IV states the results and the
closing of this door as a response. The experimental results analysis of various field tests of the system prototype,
indicate that the proposed system has the advantages such as
followed by discussions. Finally, the conclusion and future
high precision, high reliability, and controllability in response
to demands. In addition, the product prototype also has the
work are made in the last part.
economic benefits of low costs, high added-value, and other
efficient properties.
II. SYSTEM DESIGN AND THEORETICAL BASIS
Keywords-human detection; path analysis; intention analysis;
automatic accessing control system;
A. People identification by face detection
In the past ten years, the related researches on human
I. INTRODUCTION face detection can be classified into two types: the detection
method based on skin color characteristics [1] and the
Using automatic door control system is convenient and is detection method based on the training of human face
widely applied in many public places such as shores, database [2-4]. The former considers the blocks with the
businesses, transportation stations, airports, and wholesale known skin color characteristics in the images as candidate
department stores to eliminate the need of manually opening blocks, and then the existence of isosceles triangles formed
and closing the doors. However, there exists severe by eyes and mouth within the blocks can be used to confirm
drawback in the existing approaches, ex: they are unable to human faces. In the latter method, neural networks [2] or
determine the type and the intention of the detected object. In Adaboost [3-4] training methods are used to find human
other words, a puppy or a bicycle passing by will trigger an facial features and detect human faces in the established
action to open the door. In addition, even with the help of facial image database. The proposed system will use the
other technologies they may verify that the object is a real Adaboost human face detection method to determine the
person, it is not able to indentify the intention of the person. existence of human object.
For instance, a pedestrian or a person stays close to the door
chatting might still trigger the door opening/closing. B. Intention Analysis by moving trajectory tracking
Frequent false actions not only wastes control and air In order to determine whether the detected human within
conditioning energy, but also reduces equipment life time. ROI did has the intention to enter the door, the cumulative
This calls for the needs of an automatic door control system density function (cdf) obtained from tracking the trajectory
based on image recognition and intention analysis. of the human face is used to judge if the person has the
intention to enter the door. Thus, this operation will be
To overcome the above drawbacks, the action of the door divided into two parts: the tracking of human faces and the
should be based on the confirmation that the object is a estimation of probability distribution of its moving
human body, and verify his/her intention by the moving trajectory. In the first part, the tracking of human face is
trajectory to access the door. In the proposed approach, an realized by comparing if the overlapping area of human face
anti-pinch measure has been added to eliminate the false between time t and t-1 is smaller than the threshold value. In
actions among existing systems. We detect the existence of the second part, the human face moving trajectory of the
human face other than the human geometric shaped (head, person with the intention of entering will be estimated to
torso) scheme since it is well developed with reasonable figure probability distribution of x and y axis. Then this
detection rate and has the closest resemblance to real human probability distribution will be used to establish the
characteristics. In addition, the images can also be saved for probability matrices of the monitored areas to estimate the
probability values corresponding to different positions.
338
The cumulative probability of the pedestrian up to the - distribution, we use twice the standard gray scale value
th image can be obtained by (3). If the pedestrian continues 2 tAMG
1 of the Fri(t-1) anti-pinch monitoring region as the
to move toward the exit/entrance gradually or continue to threshold value [9] for foreground detection. In order to
look at the entrance, then the cumulative probability value prevent the pedestrian from repeated entering/leaving of the
Penter will exceed the threshold value THPM, then the intention specific region after entering/leaving the anti-pinch
to enter the specific region can be determined. On the monitoring region, the door closing procedure will be
contrary, if this value is smaller than the threshold value, activated with 2-second delay if there is no foreground
then it's determined that the person does not have the detected within the anti-pinch monitoring region.
intention to enter, and this case will be excluded. Since the
intention of accessing the specific exit/entrance should be The completed function blocks of the proposed system is
greater for the person closer to the exit/entrance, in this presented in figure 1 and 3.
research we assume the region with greater than average
value PM the probability matrix as the main monitored
region (MMR), and this average value of the probability III. EXPERIMENTAL RESULTS AND DISCUSSIONS
matrix PM will be used as the cumulative threshold value To evaluate the proposed system, a DM365 platform of
THPM to determine if the intention of entering exists. Thus, TI-company is selected to implement all the functions. The
when the pedestrian is walking toward the specific algorithms developed were programmed in C language and
exit/entrance location gradually, the movement trajectory of were executed under the Linux OS core. The related
the person will fall mostly in the MMR (as indicated in the parameters of the prototype system are as follows: The face
area enclosed by red area in Fig. 2(d)), and the cumulative detection rate is 0.9. The processing speed is 15 frame/sec.
probability value will become higher along with the The setup height is 2.2 meters. The angle of depression is 30
accumulation of time and with the center point of human degrees. The left and right viewing angles of the image are
face center approaching the specific exit/entrance. Once it both 63 degrees. Field tests were conducted in three different
exceeds the threshold value, it's determined that the intention locations. The content of related data is presented in table 1.
to enter exists and the door will be opened accordingly.
Otherwise, it would be determined that the intention to enter The experimental results from actual field tests are
did not exist. When the pedestrian continuously looks at the shown in table 1 while some of the life test example images
exit/entry within the MMR, Penter will exceed the threshold are shown in figures 4 and 5. Among the experiments of 125
value along with the accumulation of time. Thus, it can also trials in 3 different locations, the number of effective
be determined that the intention of entering the specific opening, within the default 2 seconds interval, is 124. The
region exists. false rejection rate was 0.008, slightly higher than the
theoretically estimated value. In the only case of failure, the
To avoid the pedestrians from being pinched by the door did open but just delayed for a while (the
closing door before entering/leaving a specific exit/entrance, reaction/response time was 4 seconds). Thus, there is
a modified temporal difference method [8] is adopted to actually zero case when a person wanted to enter but was
determine whether or not there are moving objects near the rejected. On the other hand, the number of false acceptance
exit/entrance, and the determination of whether or not the rate due to objects passing by or environmental interference
pedestrian has entered/left the specific region will be used as is zero. It is observed that the system effectively overcomes
the basis for activating the door closing procedure. In order the disadvantages of susceptibility to environmental
to confirm whether the pedestrian has entered/left the interference among existing automatic door control systems
specific exit/entrance, 1/3 imaging area in front of the while maintaining the advantages of opening the door
exit/entrance is preset to be the anti-pinch monitoring region effectively and safely.
(AMG). The temporal difference method will be used to
detect the existence of any foreground within the monitoring
region as the basis for activating the door closing procedure. IV. CONCLUSIONS
Then the temporal difference is to detect the pixel different
An intelligent automatic control system based on
from the previous moment t-1 by subtracting two adjacent
analyzing the human properties and his behavior is proposed
images in the same image sequence. When this pixel is
in this paper to improve the existing accessing control
greater than the threshold value , it will be determined as
devices. Through image analysis techniques, it is capable of
foreground pixel by
opening the automatic control door after accurate
identification of human object with the intention to enter or
1, | Fri (t ) Fri (t 1) > |
Fgi (t ) = exit from that door. This way it achieves the goals of added
0 , otherwise
(4) values such as reduction of energy consumption, improved
readiness rate of the equipment, and saving of the images of
where Fgi(t) is the foreground pixel at time t. Fri(t), Fri(t-1) people who entered/exited for future references of other
are the gray scale values of position i in the anti-pinch applications. The results of the initial field tests revealed that
monitoring region at time t and t-1 respectively. Assuming the developed prototype effectively accomplished the
that the gray scale values of images are of Gaussian
339
product design specification thus can be successfully and
practically applied to real life applications.
REFERENCES
[1] R. L. Hsu, M. Abdel-Mottaleb, and Anil K. Jain, Face Detection in
Color Images, IEEE Transactions on Pattern Analysis and Machine
Intelligence, vol. 24, no. 5, pp. 696-706, May 2002.
[2] Henry A. Rowley, Shumeet Baluja, and Takeo Kanade, Neural
Network-Based Face Detection, IEEE Transactions on Pattern
Analysis and Machine Intelligence, vol. 20, pp. 23-38, 1998.
a Fig. 1. System flowchart of (a) door opening processes, and (b) door
[3] Paul Viola, and Michael J. Jones, Robust Real-Time Face b closing processes.
Detection, International Journal of Computer Vision, vol. 57, no. 2,
pp. 137-154, 2004.
[4] Rainer Lienhart, and Jochen Maydt, An Extended Set of Haar-like
Features for Rapid Object Detection, Proceedings Of the IEEE Projection of vertical trajectory
1 Gaussian distribution
International Conference on In Image Processing, vol. 1, pp. 900-903,
Sep. 2002. 0.8
0.5
0.2
Analysis and Machine intelligence, vol. 22, no. 8, pp. 809830, 0.1
August 2000. 0
0 50 100 150 200 250
Vertical Position
340
a b Fig. 4 (a) no face is detected when the pedestrian move into the
c d ROI, (b) a face is detected at first time, (c) the face is tracked and
the probability is accumulated continuously, and (d) when the
cumulated probability is great than a threshold, the door status is
changed to open.
a b Fig. 5 Although the face is detected in (b), the door is not opened
c d since the accumulated probability is less than the threshold.
341