Aerospace and Electronic Systems - May 2019 - 54

Feature Article:

DOI. No. 10.1109/MAES.2019.2916293

Power Line Detection for Aircraft Safety Based on
Image Processing Techniques: Advances and
Recommendations
Xingchen Zhang, Gang Xiao, Shanghai Jiao Tong University, Shanghai, China
Ke Gong, East China Normal University, Shanghai, China
Ping Ye, Junhao Zhao, Shanghai Jiao Tong University, Shanghai, China

INTRODUCTION
Power line detection (PLD) has attracted increasing attention in recent years, because it has many important applications in both military and civil fields. For example, it is
vital for flight safety of helicopters. The United States
army reported that they had lost more helicopters to power
lines than against enemies in combat [1]. The Australian
Transport Safety Bureau found that 119 wire-strike accidents as well as 98 wire-strike incidents had been reported
between 1994 and 2004 [2]. One-hundred sixty nine
people were involved in these accidents, among whom
45 people were fatally injured. Besides, most aircrafts
were destroyed or damaged substantially. Examples of
helicopters crashed due to power line-strike are given in
Figure 1. For the safety of aircrafts, it is important to recognize power lines and remind pilots in advance.
In past decades, with the continuously increasing of
camera resolution, detecting power lines from aerial
images directly have become possible. Many methods
based on image processing techniques have been proposed
to detect power lines to help pilots avoid them. However,
traditional image processing-based approaches have some
shortcomings. First, features used to detect lines are
designed manually based on prior knowledge, which may
not be very effective. Second, most algorithms are
designed for straight line detection. Whereas, power lines

Authors' current addresses: X. Zhang, G. Xiao, P. Ye,
and J. Zhao, School of Aeronautics and Astronautics,
Shanghai Jiao Tong University, Shanghai 200240,
China. E-mail: (xiaogang@sjtu.edu.cn). K. Gong, East
China Normal University, Shanghai 200062, China.
Manuscript received November 29, 2018; revised April 2,
2019, and ready for publication May 10, 2019.
Review handled by P. Willett.
0885-8985/19/$26.00 ß 2019 IEEE
54

may not always be straight lines in aerial images. Third,
those algorithms are computationally expensive since preprocessing and time-consuming feature extraction are
needed. Last but not the least, a single kind of information,
such as visible image, may not be able to provide enough
information that a PLD system needs.
Consequently, a more robust, accurate, and intelligent
PLD method is needed. To utilize advantages of multimodal information, data fusion is required. Besides, with
the great success of deep learning in various fields,
researchers have begun to apply deep learning to
PLD. Considering many studies in PLD and various proposed methods, we find that it is time to conduct a thorough survey and to discuss future trends of PLD.
The remaining of this paper is organized as
follows. Section "BACKGROUND" gives some background information. Section "MAIN STUDIES ON
POWER LINE DETECTION" presents an extensive
review of current PLD methods and Section "FUTURE
PROSPECTS" proposes future prospects. Finally, Section
"CONCLUSION" concludes the paper.

BACKGROUND
POWER LINE DETECTION
A frequently used approach to help avoidance of power
lines is warning sphere, which is usually installed near airports as shown in Figure 2. Although warning spheres can
help to recognize power lines, they are difficult and timeconsuming to install. Besides, it is not possible to install
warning spheres everywhere; thus, only a very limited
part of power lines can be covered. In addition, warning
spheres may fall, which is dangerous for people beneath
them. Therefore, it is highly desirable to have a system
which can automatically detect power lines with high

IEEE A&E SYSTEMS MAGAZINE

MAY 2019

Aerospace and Electronic Systems - May 2019

Table of Contents for the Digital Edition of Aerospace and Electronic Systems - May 2019