Aerospace Science and Technology 39 (2014) 352–360
Contents lists available at ScienceDirect
Aerospace Science and Technology
www.elsevier.com/locate/aescte
Target detection approach for UAVs via improved Pigeon-inspired
Optimization and Edge Potential Function
Cong Li, Haibin Duan
∗
State Key Laboratory of Virtual Reality Technology and Systems, School of Automation Science and Electrical Engineering, Beihang University (BUAA), Beijing
100191, PR China
a r t i c l e i n f o a b s t r a c t
Article history:
Received
7 June 2014
Received
in revised form 3 September 2014
Accepted
18 October 2014
Available
online 22 October 2014
Keywords:
Unmanned
Aerial Vehicles (UAVs)
Pigeon-inspired
Optimization (PIO)
Edge
Potential Function (EPF)
Simulated
Annealing (SA)
Target
detection
In this paper, the hybrid model of Edge Potential Function (EPF) and Simulated Annealing Pigeon-
inspired
Optimization (SAPIO) algorithm is proposed to accomplish the target detection task for
Unmanned Aerial Vehicles (UAVs) at low altitude. EPF can be calculated from the edge map of the
original image and provide a kind of attractive pattern for the given target, which is conventionally
exploited by the optimization algorithms. Pigeon-inspired Optimization (PIO) is a novel bio-inspired
computation algorithm, which was inspired from the homing characteristics of pigeons. In this paper,
the simulated annealing mechanism is adopted in our SAPIO algorithm for maximizing the value of EPF.
A series of comparative experiments with standard Genetic Algorithm (GA), Particle Swarm Optimization
(PSO), Artificial Bee Colony Optimization (ABC) and PIO algorithms demonstrate the robustness and
effectiveness of our SAPIO algorithm. Meanwhile, the proposed approach can guarantee accurate target
matching.
© 2014 Elsevier Masson SAS. All rights reserved.
1. Introduction
Compared with manned aircraft, Unmanned Aerial Vehicles
(UAVs) are affordable and convenient for high-risk missions. To-
day,
UAV has been exploited to perform special missions carrying
some important equipment such as GPS, optic camera and vari-
ous
sensors [21]. While performing specific tasks, the UAV is an
efficient tool because of its superior maneuverability and strong
viability [14]. Furthermore, due to the rapid development of artifi-
cial
intelligence technology, the UAV technology plays a vital role
in the technology field of the nation and is essential for improving
the security of the society [7,8].
As
image sensors have become more and more advanced, it has
become imperative to design a satisfying target recognition system
for UAVs in order to achieve autonomous reconnaissance and de-
tection.
The target recognition and detection method for UAVs has
been investigated quite intensively in recent years. Ding et al. [5]
used
the template matching method to recognize and track run-
way
in the image sequences. Deng and Duan [4] proposed a novel
biologically inspired model via improved artificial bee colony and
visual attention to perform edge detection. Niu et al. [16] exploit
target regions in DWT domain to perform the infrared and visible
*
Corresponding author. Tel.: +86 10 8231 7318.
E-mail
address: hbduan@buaa.edu.cn (H. Duan).
image fusion, which can make UAV realize environment and per-
form
detection tasks efficiently. Meanwhile, shape representation
and matching methods such as Hausdorff distance matching [9]
and
Charmfer matching [2] are key points in all sorts of ways, and
have been extensively adopted for target detection and recognition
problems [22].
Energy
Potential Function (EPF), firstly proposed by Dao et al., is
a
novel approach for the edge-based detection in digital images [3].
Similar with the potential generated by the electrostatic field, EPF
is exploited to model the potential generated by edge structures
of the image. However, EPF is a multi-dimensional and complex
function, which is hard to optimize.
In
recent years, Evolutionary Algorithms (EAs) including Particle
Swarm Optimization (PSO) [17], Artificial Bee Colony Optimization
(ABC) [11], and Genetic Algorithm (GA) [10] have become very
popular in the optimization community and successfully applied
to a wide range of problems. Mirghasemi et al. [15] exploited the
combination of the PSO and FCM to solve the sea target detec-
tion
problem. Wang and Duan utilized the improved version of
biogeography-based optimization (BBO) for unmanned helicopter
formation [20,18]. Tao et al. used ant colony optimization algo-
rithm
and fuzzy entropy for object segmentation [19].
Pigeon-inspired
Optimization (PIO) algorithm is a novel swarm
intelligence optimization algorithm, which was firstly invented by
Duan in 2014 [6]. Motivated by the homing characteristics of pi-
geons,
two operators including map and compass operator and
http://dx.doi.org/10.1016/j.ast.2014.10.007
1270-9638/
© 2014 Elsevier Masson SAS. All rights reserved.