978-1-4799-7492-4/15/$31.00 © 2015 IEEE
Threshold Image Segmentation Based on
Dynamic Mutation and Background Cooperation
Abstract—Quantum-behaved particle swarm optimization
(QPSO) algorithm simulates quantum mechanics among
individuals. For improving the local search ability of QPSO and
guiding the search, an improved QPSO algorithm based on
combining the dynamic mutation and cooperative background
(MCQPSO) is proposed in this paper. The dynamic Cauchy
mutation strategy is introduced to enhance the global search
ability. The cooperative background strategy is used to change
the updating mode of the particles in order to guarantee the
effectiveness and simplification. The MCQPSO algorithm keeps
the diversity of the population, and increasing convergence rates.
Results compared with some previous study show that
the
MCQPSO algorithm performs much better than the Sun
Jun’s
Cooperative Quantum-Behaved Particle Swarm Optimization
(sunCQPSO) and WQPSO algorithm in terms of the image
segmentation accuracy and the computation efficiency.
Keywords—Particle swarm optimization, Quantum-Behaved
Particle Swarm Optimization, quantum space, mutation,
cooperative, image segmentation
I. INTRODUCTION
A. Background of Evolutionary Algorithm and Image
Segmentation
Evolutionary algorithm is a new evolutionary computation
technology; especially the evolution strategy and the genetic
algorithm have a very special connection. The particle swarm
optimization (PSO) algorithm is a kind of evolutionary
computation technology from the birds feeding behaviors.
The past decades saw the obvious progress of PSO in the field
of evolutionary algorithm. Particle swarm optimization (PSO)
is a kind of optimization tool which is based on iteration [1,
21],
PSO is easy to understand. With less parameter, it is not
difficult to implement and program. In [10], we proposed a
Quantum-based particle swarm optimization combined
classical QPSO with cooperative quantum (CQPSO). The
general PSO algorithm can't search the global optimal
solution with probability 1, in order to enhance the search
ability of PSO, many improvements have been made over
years. So PSO algorithm is extended to the quantum space [2,
3], the extension of the PSO algorithm to quantum space has
introduced the optimization algorithm in which the particles
have the behavior of quantum particle swarm, namely the
QPSO algorithm. Compared with the PSO algorithm, the
QPSO algorithm does not need the particle's velocity
information, so the control parameter is less and the global
search ability has been enhanced [4]. But it will appear the
situation of the local convergence, so the follow-up for the
QPSO algorithm with a series of improved algorithm are put
forward.
Segmentation technology plays a very important role in
image processing, in this method; the whole image is
segmented into number of distinct, the same nature regions
preserving some defined properties. Therefore, the
segmentation technology applied to the medical and other
technical field, and effective medical image processing
methods help the doctor to obtain more important information.
Medical images processing technologies include threshold
segmentation, nuclear magnetic resonance (NMR) technology,
computed tomography (CT), and single photon emission
tomography camera technique. Thresholding is referred to as
a popular method for image segmentation, which is used to
separate image with its background. After segmentation,
images can be used in clinical operation guide and local bulk
effect correction.
B. Present Medical Image Segmentation and the
QPSO Algorithm
The medical image has the characteristics of complexity
and diversity, so the imaging principle of medical image and
the general image is different. Because the influence of noise
and the local effect, so the medical image is used more widely
in technology research [2]; In order to solve the problems of
medical image segmentation, there are many segmentation
algorithms that have been put forward successfully, such as:
the wavelet theory, clustering, double clustering method,
genetic algorithm and multi-resolution, etc. However, the
medical image segmentation research has its remarkable
characteristics. Firstly, medical image segmentation is applied
to clinical with high accuracy. Secondly, the segmentation
technology has multi-dimensions and 2 dimensions
This work was supported by the Program for New Century Excellent
Talents in University of China (No. NCET-12-0920), the Program for New
Scientific and Technological Star of Shaanxi Province of China (No.
2014KJXX-45), the National Natural Science Foundation of China (Nos.
61272279, 61272282, 61371201, and 61203303), the Fundamental
Research Funds for the Central Universities of China (Nos. K5051302049,
K5051302023, K50511020011, K5051302002 and K5051302028), the
Provincial Natural Science Foundation of Shaanxi of China (No.
2011JQ8020), the Fund for Foreign Scholars in University Research and
Teaching Programs of China (the 111 Project) (No. B07048), and EU
IRSES project (No. 247619)
Yangyang Li, Member, IEEE, Yang Yue, Licheng Jiao, Senior Member, IEEE and Ruochen Liu,
Member, IEEE
Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China,
International Research Center for Intelligent Perception and Computation, Xidian University, Xi'an 710071, China
yyli@xidian.edu.cn
3271