Digital Signal Processing Reference
In-Depth Information
particularity color of traffic signs. The main colors adopted in traffic sign system are
blue, yellow, red, and black. Examples of traffic sings from Chinese national standard
GB5768-1999 database are shown in Fig.1.
Fig. 1.
Examples of traffic signs in Chinese national standard GB5768-1999 database
Taking into account the specificity of the color of road signs, for the first time in
this paper, traffic sign feature extracting based on Pulse Coupled Neural Network
(PCNN) in different color space are investigated. Each entropy series of traffic images
in
R, G, B, H, S, V
color space is extracted as feature vector for recognition, to find
out the optimal color space for feature extraction.
The paper is organized as follows. Section 2 addresses the theory of PCNN. Sec-
tion 3 addresses the traffic images and entropy series of images. Section 4 reports the
experimental results. Finally, Section 5 summarizes the paper.
2
Description of PCNN Model
In image processing, PCNN model is organized as a one lever 2-D network, which has
the same size with the input image. Each neuron is corresponding with a pixel in the
image. The basic PCNN neuron is shown in Fig.2. There are three components in the
neuron: input part, linking park, pulse generator. The PCNN neuron accepts the feed-
ing input and the linking input, and then generates the internal activity. Feeding input is
the basic input from the neuron's, which consists of the neighboring pixels of corres-
ponding pixels in the input image. Linking input is the input of lateral connections with
neighboring neurons. After compared to a dynamic threshold, Feeding input and Link-
ing input are then combined to create the internal state.
For the convenience of simulation, the PCNN is described as iteration equations as
follows [7]:
−
α
Fn
[]
=
e Fn
[
−
1]
+
V
M Yn
[
−
1]
+
S
F
(1)
ij
ij
F
ijkl
ij
ij
kl
Ln e Ln
[]
=
−
α
[
−
1]
+
V
WYn
[
−
1]
L
(2)
ij
ij
L
ijkl
ij
kl
Un Fn
[]
=
[](1
+
β
Ln
[])
(3)
ij
ij
ij
−
=− +
α
Tn Tn
[]
[
1]
e
VYn
[]
T
(4)
ij
ij
T
ij
1[ ]
Un Tn
>
[ ]
ij
ij
Yn
[]
=
(5)
ij
0
Un Tn
[ ]
≤
[ ]
ij
ij
