Pattern Recognition 60 (2016) 106–120

Contents lists available at ScienceDirect

Pattern Recognition
journal homepage: www.elsevier.com/locate/pr

Learning feature fusion strategies for various image types to detect

salient objects
Muhammad Iqbal n, Syed S. Naqvi, Will N. Browne, Christopher Hollitt, Mengjie Zhang
School of Engineering and Computer Science, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand

art ic l e i nf o a b s t r a c t

Article history: Salient object detection is the task of automatically localizing objects of interests in a scene by sup-
Received 7 March 2016 pressing the background information, which facilitates various machine vision applications such as
Received in revised form object segmentation, recognition and tracking. Combining features from different feature-modalities has
20 April 2016
been demonstrated to enhance the performance of saliency prediction algorithms and different feature
Accepted 4 May 2016
Available online 21 May 2016
combinations are often suited to different types of images. However, existing saliency learning techni-
ques attempt to apply a single feature combination across all image types and thus lose generalization in
Keywords: the test phase when considering unseen images. Learning classifier systems (LCSs) are an evolutionary
Object Detection machine learning technique that evolve a set of rules, based on a niched genetic reproduction, which
Saliency Map
collectively solve the problem. It is hypothesized that the LCS technique has the ability to autonomously
Learning Classifier Systems
learn different feature combinations for different image types. Hence, this paper further investigates the
XCS
Pattern Recognition application of LCS for learning image dependent feature fusion strategies for the task of salient object
detection. The obtained results show that the proposed method outperforms, through evolving gen-
eralized rules to compute saliency maps, the individual feature based methods and seven combinatorial
techniques in detecting salient objects from three well known benchmark datasets of various types and
difficulty levels.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction detection concentrate on constructing deterministic tailor-made

Visual saliency has recently attracted much computer vision
research, giving birth to a new sub-domain known as salient ob-
ject detection [1]. For salient object detection, the task is to detect
the salient, attention grabbing object(s) in a scene and subse-
quently segment it in its entirety [2,3]. It is similar to the problem
of figure-ground segmentation [4–6], but differs from the tradi-
tional segmentation problem as the task is simply to find the most
salient object rather than completely partitioning the image into
perceptually homogeneous regions [7]. Salient object detection is
actually the task of marking regions of interest in a scene, which
facilitates various computer vision applications, e.g. image seg-
mentation [8], image retrieval [9,10], picture collage [11,12], object
recognition [13] or image compression [14].
Most methods specialized for the task of salient object
n
Corresponding author.
E-mail addresses: muhammad.iqbal@ecs.vuw.ac.nz (M. Iqbal),
syed.saud.naqvi@ecs.vuw.ac.nz (S.S. Naqvi),
will.browne@ecs.vuw.ac.nz (W.N. Browne),
christopher.hollitt@ecs.vuw.ac.nz (C. Hollitt),
mengjie.zhang@ecs.vuw.ac.nz (M. Zhang).
features [15,16] such as color or color gradient and apply heur-
istics to combine them. A class of models [17–19] use low, mid
and high-level features to learn a single set of weighting para-
meters for combining features, but apply them across multiple
types of images, e.g. images with cluttered backgrounds or
multiple objects of interest. Therefore, such techniques in-
herently lose generalization when operated on test sets with
different images having various properties and sets of features.
An alternative approach is to learn model parameters using an
assembly of weak learners, which increase generalization. How-
ever the quality of final solution depends upon the performance
of individual learners and can be degraded if one of the learners
is not optimal [20].
A learning classifier system (LCS) is a rule-based machine
learning technique in which each rule relates sections of the
feature space with a classification and a measure of accuracy
[21,22]. To address the issue of loss in generalization on unseen
image types and to make the system general for all image types,
previously we utilized the strength of LCS to autonomously di-
vide the feature space into niches and construct rules covering
each image type [23]. The aim of this paper is to extend and

http://dx.doi.org/10.1016/j.patcog.2016.05.020
0031-3203/& 2016 Elsevier Ltd. All rights reserved.
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
demonstrate the LCS technique proposed in [23] by fully in-
vestigating niching of image types and demonstrating perfor-
mance on a wide range of domains and salient object detection
benchmark techniques.
The rest of the paper is organized as follows. Section 2 briefly
describes the related work in salient object detection. In Section 3
the proposed LCS technique to detect salient objects in an image is
detailed. Section 4 introduces the datasets, parameter settings, and
performance measures used in the experimentation. In Section 5
experimental results are presented and compared with existing
state-of-the-art systems. Section 6 provides an analysis of the
evolved classifier rules obtained using the proposed LCS system. In
the ending section this work is concluded and the future work is
outlined.
2. Background
This section introduces the necessary background in learning
classifier systems, and the related work in salient object detection.
2.1. Learning classifier systems
Traditionally, an LCS represents a rule-based agent that in-
corporates a genetic algorithm (GA) and machine learning to solve
a given task by evolving a population of interpretable classifiers.
Each classifier covers a part of the feature space that may be
overlapped with other classifiers. The LCS technique has been
successfully applied to a wide range of problems including clas-
sification, data mining, control, modeling, image processing and
Fig. 1. Overview of a learning classifier system [28].
107
optimization problems [24–31].
The proposed LCS method, to be presented in this study, to
compute saliency maps is based on XCS [32], which is a well-
tested LCS model. In XCS, the learning agent evolves a population
[P ] of classifiers, as depicted in Fig. 1, where each classifier con-
sists of a rule and a set of associated parameters estimating the
quality of the rule. Each rule is of the form ‘if condition then ac-
tion’, where condition is used to match input observations, and
the corresponding action predict the class label for a given ob-
servation. Commonly, the condition in a rule is represented by a
conjunction of predicates using one predicate for each corre-
sponding input feature; and the action is represented by a nu-
meric constant.
In XCS, on receiving the environmental input state s, a match
set [M ] is formed consisting of the classifiers from the population
[P ] that have conditions matching the input s. For every action ai in
the set of all possible actions, if ai is not represented in [M ] then a
covering classifier is randomly generated. After that an action a is
selected to be performed on the environment and an action set [A]
is formed, which consists of the classifiers in [M ] that advocate a.
After receiving an environmental reward, the associated para-
meters of all classifiers in [A] are updated. When appropriate, new
classifiers are produced using an evolutionary mechanism, usually
a GA. Additionally, in XCS overly specific classifiers may be sub-
sumed by any more general and accurate classifiers in order to
reduce the number of classifiers in the final population [33]. For a
complete description, the interested reader is referred to the ori-
ginal XCS papers by Wilson [32,34], and to the algorithmic details
by Butz and Wilson [35].
2.2. Salient object detection
Visual attention is a fundamental research problem in psy-
chology, neuroscience, and computer vision literature. Researchers
have built computational models of visual attention to predict
where humans are likely to fixate [36]. Recently, this work has
been expanded to identify salient objects in a scene for object
detection and localization. Salient object detection is a difficult
problem in computer vision as natural scenes can include objects
with cluttered backgrounds (making it difficult to distinguish the
object from background based on its features) and scenes con-
taining multiple objects.
Deterministic methods to detect salient objects include fine
human-constructed features, but they usually combine them lin-
early, thus neglecting the importance of individual features [16].
Machine learning approaches have the ability to learn feature
importance during combination, which enhances their perfor-
mance in challenging cases such as scenes with cluttered back-
grounds and multiple objects [37].
Tong et al. [38] used 73 texture and color features by exploring
both global and local cues to compute a saliency map. However,
the simplistic nature of feature combination (i.e., the average of
the local and global features) compromises the final saliency
output on difficult cases of saliency detection. Judd et al. [17]
learned a model of saliency from 33 features (including low, mid
and high level features) to predict human eye fixations. They used
support vector machines (SVMs) with linear kernels to learn fea-
ture weightings, while Zhao and Koch [19] used least square re-
gression to learn weights for eye fixation prediction using basic
saliency features (i.e., color, intensity and orientation). Both the
discriminative approaches lose generalization on a subset of
images due to a single weighting scheme being applied to features
for all image types. Singh et al. [39] applied a constrained Particle
108 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Swarm Optimization method to determine an optimal weight
vector to combine different features to obtain a saliency map. The
single solution set learned after the evolutionary process hinders
the generalization performance of their method on the large set of
testing images.
The recent discriminative regional feature integration (DRFI)
work of Jiang et al. [7] employs regression to automatically select
and integrate features. The inclusion of high dimensional features
for learning a regression model enables their approach to dis-
cover discriminative features and achieve robust performance on
unseen data. However, computing high dimensional features and
running a regressor for each image to increase generalization
comes at a cost of additional computational time and limits
scalability with increasing number of regions per image. Despite
the inclusion of multiple segmentations at different scales, the
region based saliency computation is prone to inherent limita-
tions, such as inappropriate annotation and non-uniform saliency
assignment. Additionally, in scenarios where feature perfor-
mance is heavily dependent upon important feature related
parameters, joint optimization of such parameters cannot be
easily incorporated into the automatic feature integration
process.
Moreover, related work on alternating optimization for multi-
view varied feature fusion has been proposed in past works that
takes into account the complementary information of multiview
data [40,41].
All the above mentioned approaches achieve reasonably good
results, however they lose generalization capability as they only
learn a single set of weights to combine features for all image
types. AdaBoost [18,20] learns the task of salient object detection
using an assembly of weak learners (hence increasing general-
ization). However the quality of final solution depends upon the
performance of individual learners and can be affected drastically
by one of the learners in the decision tree, badly affecting the
generalization of the overall system [42]. Once again, AdaBoost
does not divide the feature space into niches depending upon
image types, which affects generalization on unseen images. Due
to cooperative nature of the evolved rules, the LCS technique has
an inherent capability to autonomously divide the feature space
into niches. Previously we adapted a supervised classifier system,
known as XCS with Computed Action (XCSCA) [43], to learn dif-
ferent combinations of image features and construct rules cov-
ering each image type [23]. This study substantiates that work by
fully investigating the niching of different image types and de-
monstrating performance on a wide range of domains and salient
object detection benchmark techniques.
3. Salient object detection using learning classifier systems
Commonly, salient objects in an image are detected by determining
a saliency map for the input image, which is a computed image that
attempts to emphasize the object(s) to be detected. In a supervised
learning system, it is necessary to provide the target saliency map (i.e.,
the ground truth) along with the input image during training. A ground
truth is a manually segmented binary image that emphasizes the ob-
ject to be detected. During the testing process, only the image features
computed using the input image are provided to the system. The
training and testing processes used in the proposed XCSCA-based ap-
proach to detecting salient objects are briefly described in Algorithms
1 and 2, respectively.
Algorithm 1. The training process in the proposed approach. Here
maxProbs denotes the maximum number of training instances,
which is usually greater than the number of images in the training
set S because a single image can be used more than once as a
training instance.
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Algorithm 2. The testing process in the proposed approach.
The remaining of this section explains the new methods in-
corporated in XCSCA to compute saliency maps in an effort to
detect salient objects from various types of images by learning
different feature fusion strategies. The new extensions introduced
in the proposed technique are: the design of input features, the
mechanism to match an input image with a population of classi-
fiers, the mechanism to compute actions that are essentially the
saliency maps, and the error function to calculate an error be-
tween a computed saliency map and the target saliency map.
3.1. Design of input features
Instead of matching the input image at pixel-level against
conditions of classifier rules, we will compute the following nine
saliency-based features for each input image. Each of this produces
a two-dimensional real-valued arrays. We carefully select potential
features from previous work [44,45,8,46,47] that are suited for the
task of salient object detection and have been previously shown to
correlate with visual attention in different experimental settings.
To thoroughly evaluate the learning performance of models, we
have chosen features that complement each other well, with each
one performing better than others for a particular image type in
past experiences.
 f 0: A global feature that assigns low saliency to colors that vary
a lot in the spatial domain, which is based on the work of Liu
et al. [1]. To compute the spatial variance of colors in an image,
all the colors are modeled by Gaussian Mixture Models (GMM).
Afterwards, each pixel in the spatial domain is assigned to a
109
color component of the GMM. Next, horizontal and vertical
variances of each color component of the GMM are calculated
and added to obtain the total variations of colors in the spatial
domain. Finally, the colors having high total variance are
assigned low saliency, while the colors exhibiting low variance
in the spatial domain are assigned high saliency values.
 f 1: A global feature that captures the contrast between clusters
obtained through k-means segmentation, inspired by the work
of Fu et al. [45]. The contrast cue of a cluster is computed by
accumulating its distance to all other clusters.
 f 2: A global feature computing the spatial distribution of pixels
in a cluster with respect to the image center, inspired by the
work of Fu et al. [45]. The global spatial distribution of pixels in
a cluster from the image center is measured as the mean of
Euclidean distance between pixels in a cluster to the image
center.
 f 3: A region based feature that computes the global contrast
between spatial neighboring regions only [8]. The image is first
segmented using graph-based image segmentation. Next, a
quantized color histogram is constructed for each region. After-
wards, saliency for each region is computed as its weighted
color contrast to all other regions in the image. The weights are
the number of pixels contained by a region to emphasize
contrast to larger regions, while the contrast itself is measured
by the color distance metric between the two regions.
 f 4, f 5: Two low-level region-based color features adapted from
the work of Naqvi et al. [47]. One color feature for each region is
computed by accumulating the earth mover's distance (EMD) of
its LAB histogram from the histograms of all other regions in the
image, while the other is computed by measuring EMD between
the histograms of a region and its neighboring regions only.
Additionally, contrast from the boundary regions is also
exploited in these features to enhance their discriminative
110 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120

power.
 f 6: A mid-level feature that uses the objectness of image win-
dows to highlight salient objects, based on the work of Alexe
et al. [46]. The objectness measure for a window is based on
four image cues. The first cue is multi-scale image saliency
based on the work of Hou and Zhang [48]; the second cue is the
color contrast of a window from its surrounding regions,
computed as the Chi-square distance between the Lab histo-
grams of the window and its surrounding super-pixels; the
third cue is computed by measuring the edge density inside a
window; finally, the last cue counts the number of super-pixels
that have their pixels both inside and outside the boundary of
the window. The number of such super-pixels that have their
pixels both inside and outside the window boundary must be
low for a window having a high probability of containing the
object(s).
 f 7: A feature that groups regions based on their objectness score.
Similar regions in terms of objectness scores are merged to form
a larger region. For each region, its difference from all other
regions is computed in terms of objectness scores to form a
difference matrix of size equal to the square of the number of
regions. From the difference matrix a global threshold is cal-
culated by finding the smallest difference that exist between
neighbors. Afterwards, a local process compares regions only
with their neighboring regions and groups those having a dif-
ference less than the global threshold by assigning them the
same objectness.
 f 8: A feature that highlights salient patterns based on the work
of Naqvi et al. [47]. The salient patterns are determined by
finding any outstanding patches that have a large distance from
neighboring patches. Match distance is employed to compute
the histogram distance between patches due to its ability to
capture cross-bin similarities/dissimilarities. In addition, intra-
patch variance is exploited for computational efficiency.

3.2. Input matching scheme

If classifier conditions in an LCS are matched directly to the

Fig. 2. The novel encoding scheme to match an input image against the classifier
population in order to enable generalization in classifier rules.
computed image features, then it will be hard to evolve any gen-
eralized classifier rules due to the large-sized two-dimensional
image features [26]. Therefore, in order to enable generalization,
we introduce a novel encoding scheme to match an input image

Fig. 3. From left to right: image, ground truth, features f 0 − f 8 in the same order as listed in Section 3.

Table 1
The learning parameters used in the proposed method.

Parameter Description Value

FI Initial fitness value of a new classifier 0.01

α Fitness fall-off rate used to calculate fitness of a classifier 0.1
β Learning rate used to update fitness of a classifier 0.2
ϵ0 Error threshold in accuracy under which a classifier is considered to be accurate 10
ν Fitness exponent used to calculate fitness of a classifier 5
θGA Threshold above which the GA is applied in a match set 25
χ Probability of applying (two-point) crossover in the GA to produce two offspring 0.8
μ Probability of mutating an attribute in an offspring 0.01
θdel Experience threshold above which fitness of a classifier may be considered in its probability of deletion 20
δ Fraction of mean fitness of population below which fitness of a classifier may be considered in its probability of deletion 0.1
θsub Experience threshold above which a classifier may subsume another classifier 20
Fr Fitness reduction factor used to reduce the average fitness of two parent classifiers before assigning it to offspring 0.1
r0 A parameter used in the covering operation to determine the distribution range of the spread in each interval 0.7
m0 A parameter used in the mutation operation to modify the distribution range of intervals in offspring 0.5
x0 A constant input parameter used to compute action of a classifier 0.5
δrls A scaling factor used to initialize the covariance matrix in a classifier 100
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120 111

Fig. 4. Comparison of individual features with XCSCA on all datasets in terms of PR curves. From top row to bottom row: MSRA, SOD, SED2 and average PR curve on all
datasets.
112 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Fig. 5. Comparison of individual features with the proposed XCSCA method in terms of segmentation based measures, i.e. F-measure, precision and recall. (a) MSRA, (b) SOD,
(c) SED2 and (d) average measures on all datasets.
against classifier conditions. In this encoding scheme, each com-
puted image feature fi will be encoded as a real-valued constant di
to be matched against classifier conditions. The real-valued con-
stant for a feature is computed using earth mover's distance (EMD)
[49] from a two-dimensional artificial feature consisting of all
ones. EMD is a way of converting an image into a number, which is
defined as the minimum cost for transforming one histogram into
the other such that there exists a ground distance between the
features.1 Consequently, conditions in classifier rules will be en-
coded as a concatenation of real-valued intervals so that the sys-
tem can evolve generalized classifiers. To aid in understanding the
novel encoding scheme, the classifier matching mechanism is
depicted in Fig. 2.
If the current input state is not matched by any classifier in the
population during the training phase, a random classifier is cre-
ated to cover the current input. However, if an input is not mat-
ched during the testing phase, we compute Euclidean distance of
each classifier condition from the current input; and consider the
1
We used the fast implementation of EMD with thresholded ground distances
based on the work of Pele and Werman [50].
closer “ |P | × tournamentSize ” classifiers to be matched, where |P | is
the number of classifiers in the population and tournamentSize is a
constant parameter, usually chosen as 0.4.
3.3. Computing actions/saliency maps
In an LCS, usually, the action in a classifier rule is represented
by a fixed scalar value. However, to learn a task that has a large
number of classes, it is beneficial to use a mechanism to compute
actions instead of using fixed scalar actions in classifier rules
[43,28].
In this work, the action (the saliency map) in a classifier rule is
computed as a linear function of the input image and a weight
vector w, similar to XCSCA [43]. In this function, the computed
two-dimensional features are linearly combined with the evolved
weights to produce the required saliency map for the matched
input image, as given in Eq. 1. Here x0 denotes a constant input
parameter. The weights are evolved using recursive least squares,
as described in [51].
SaliencyMap = w0 x 0 + ∑ wi + 1fi ; i = 0, 1, 2…8. (1)
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120 113

Fig. 6. Comparison of methods with respect to PR curves. From top row to bottom row: MSRA, SOD, SED2 and average PR curve on all datasets.
114 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120

Fig. 7. Comparison of methods with respect to segmentation based measures, i.e. F-measure, Precision and Recall. (a) MSRA, (b) SOD, (c) SED2 and (d) average measures on
all datasets. The methods are sorted according to their F-measure scores.

To form a match set we use the real-valued distance (denoted
by d) for each of the two-dimensional features, but to compute the
saliency map the corresponding feature itself is used.
3.4. Error function
4. Experiment design
This section describes the data sets, parameter settings, per-
form measures, and the state-of-the-art methods used in this
study to compare the results of the proposed approach XCSCA.

The goal of the proposed approach is to evolve saliency maps
for each input image such that the error between the computed
saliency map and the target saliency map is minimized. The error
is calculated using an error function that determines the difference
between the computed saliency map and the ground truth of the
input image. The calculated error is used to update the associated
parameters in the corresponding classifiers. The error function
used in this work is:
TP + TN
Error = 1 − , classify multiple objects as one by placing them in a single rec-
TP + FP + TN + FN
where TP and TN are the number of correctly classified positive
and negative pixel values (for a single image) between the com-
puted saliency map and the ground truth. FP and FN represent the
falsely classified pixel values.
4.1. Data sets
This work employs the following commonly used benchmark
data sets in the field of computer vision: Microsoft Research Asia
(MSRA) [44], Salient Object Dataset (SOD) [52], and Segmentation
Evaluation Dataset (SED2) [53].
The MSRA data set is comprised of 25,000 images in total and
includes ground truth annotations in the form of labeled rec-
tangles from multiple users. These ground truth annotations
(2)
tangle and also do not cater for pixel-wise accuracy. To remedy
these effects, a set of 1000 images were manually segmented by a
single user to obtain binary masks [54]. These ground truth masks
consider the effect of pixel-wise accuracy and multiple objects and
is widely accepted in the field of computer vision as a standard
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Fig. 8. Visual comparison of selected saliency methods on representative images from MSRA (test images), SOD and SED2 datasets.
benchmark for saliency evaluation. All images and respective
ground truth are resized to 200  200 to leverage computational
efficiency. Consequently the computed feature maps are of the
same size. Fig. 3 shows some representative images with their
ground truth and respective features from three data sets. It can be
observed that no single feature accurately captures the ground
truth.
The SOD [52] benchmark contains 300 images that are difficult
for salient object detection due to the cluttered backgrounds and
ambiguous salient object scenes. Boundary level ground truth in-
formation is available for all images. The SED2 [53] contains 100
two-object images along with pixel-wise ground truth and is a
subset of the segmentation evaluation database.
4.2. Parameter settings
For all the experiments conducted in this study, the number of
classifiers used is 2000 and the number of training instances is
20,000. The selection method used to select two parent classifiers
in the GA is tournament selection with tournament size ratio 0.4.
GA subsumption is activated whereas action set subsumption is
deactivated. The remaining learning parameters and their values
used in this study are shown in Table 1, which are commonly used
in the literature, as suggested by Butz and Wilson [35].
The training has been conducted on the MSRA data set only as
it includes images of different types, e.g. cluttered background,
multiple objects, large salient objects, small salient objects, and
115
faces, persons, objects and text. The resulting classifier populations
are tested on each of the three data sets. All the experiments have
been repeated 30 times with a different seed in each run, where
the MSRA data set has been randomly divided into a training set of
700 and a test set of 300 images in each experiment. Each result
reported in this work is average of the 30 runs.
4.3. Performance measures
The performance is measured using precision–recall (PR)
curves and the segmentation quality. The PR curves are drawn by
computing 256 pairs of average precision and recall values. To
compute the 256 pairs, the saliency maps are thresholded using
256 thresholds in the range 0 to 255 to generate 256 binary maps.
The binary maps are compared with the corresponding ground
truth map for each image to get 256 pairs of precision and recall
values for each image. The average of precision recall pairs over
the whole dataset gives the final PR curve. The increase of recall
values on the x-axis of the PR curves corresponds to decrease of
threshold values from 255 to 0.
We present the original (i.e. empirical) PR curves as well as inter-
polated PR curves in the results section. The interpolated PR curves are
computed by the method “maximum precision for each recall” [55],
which ensures the monotonically decreasing trend and has been
agreed upon by researchers in the document retrieval field.
In order to compare the segmentation quality of different
methods, the saliency maps are thresholded adaptively based on
116 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120

Table 2
A sample of the experienced and accurate classifier rules, obtained in a typical run, using the proposed LCS system. The names of images covered (i.e. matched) by a rule are
listed under the rule.

No. Condition Weight Vector

1 [0.00 1.00] [0.03 0.68] [0.00 1.00] [0.22 1.00] [0.00 0.89] [0.00 0.44] [0.65 1.00] [0.02 0.66] 416.2, 352.9,183.8, 268.0,1111.5, 141.0,701.9,752.8,1135.5,397.1
[0.00 0.43]
0_13_13700, 0_18_18530, 0_24_24558, 0_2_2301, 0_6_6646, 10_268015474_2bd515353c, 10_52202431.dscn0209.png ,10_59242003.img_4193, 1_55_55005,
1_57_57848

2 [0.00 0.47] [0.00 0.80] [0.00 0.74] [0.00 0.82] [0.00 0.43] [0.00 0.27] [0.99 1.00] [0.00 0.77] 400.5, 72.7,980.0,49.6,272.7,141.5,1311.2,294.8, 110.6,628.8
[0.00 1.00]
0_18_18530, 0_19_19593, 0_21_21781, 0_2_2301, 0_7_7821, 10_260372546_03d18a4e9e, 10_267183829_e8538b16ff.png ,10_268122508_c361b4db6c,
10_268252475_3716432836_m, 10_268262571_a713359657, 10_41105485.05_0226_aabbspoth2, 10_43642602.pasqueflowerpasqueflowerearlyeve2,
10_49513258.0509180006.fsdm, 10_52202430.dscn0206, 10_52202431.dscn0209, 10_54628916.jan8_06_536, 10_54628917.jan8_06_537, 10_54628920.
jan8_06_540, 10_66830630.sejklgec.sept10_06_746, 1_26_26013, 1_51_51078, 1_55_55005, 1_57_57848, 1_58_58671, 1_61_61269, 1_65_65984

3 [0.00 1.00] [0.20 0.90] [0.40 1.00] [0.00 0.69] [0.00 0.93] [0.00 0.52] [0.15 1.00] [0.00 0.79] 559.7,255.1,517.9,296.3,15.9,472.2,539.3,40.2,627.7,787.9
[0.00,0.70]
0_11_11164, 0_11_11981, 0_12_12891, 0_13_13347, 0_13_13386, 0_13_13700, 0_15_15022, 0_16_16704, 0_16_16768, 0_18_18565, 0_19_19593, 0_1_1288, 0_1_1409,
0_1_1427, 0_21_21244, 0_21_21394, 0_24_24453, 0_2_2756, 0_4_4240, 0_5_5091, 0_5_5291, 0_6_6646, 0_7_7478, 0_7_7917, 0_8_8859, 0_9_9398, 0_9_9453,
10_00000069_018, 10_00000093_007, 10_144439584_97e8823d39, 10_1557631_c5d89caa41, 10_156234186_1b16bc540f, 10_161208056_248b2c2ab6,
10_224830165_1fc5dcfecf, 10_236916255_307bc2272c, 10_262725611_c3ce2b827e, 10_264977628_cde9f779bc, 10_266774040_f40061481f,
10_267183829_e8538b16ff, 10_267412727_1727822888, 10_267744690_ac99310c04, 10_267780676_e2346e581a, 10_267781224_2039c2b3fb,
10_267781788_2765356aeb, 10_267935839_30e321dbe5, 10_267975486_b9cd08a46f, 10_267976432_b4e9429cff, 10_268001846_db119074c8,
10_268014760_9676395938, 10_268129990_04e858de05, 10_268168010_d3852a2e6e, 10_268211019_270132655c, 10_268217391_a99ec26edb,
10_268231155_a6210a13b7, 10_268231902_1586ec90e2, 10_268252475_3716432836_m, 10_29912713.stream5sm, 10_36687635.gpzoo_przewalski, 10_38173647.
pbcats06, 10_40677597.img_1543, 10_43047581.ds20050506_0127awfcat, 10_44859877.10, 10_52202392.dscn0140, 10_52202445.dscn0226, 10_52202453.
dscn0243, 10_52202457.dscn0251, 10_52202466.dscn0266, 10_58229768.20060405022_object_of_fancy, 10_59242003.img_4193, 10_59271038.138_3822,
10_59271075.img_1787, 10_66830586.6whtssec.sept10_06_705, 10_96835757_c609dbaa80, 10_98048098_af08c0f2df, 1_26_26013, 1_26_26532, 1_27_27034,
1_35_35795, 1_43_43206, 1_47_47611, 1_47_47818, 10_66830586.6whtssec.sept10_06_705, 1_51_51078, 1_53_53496, 1_54_54098, 1_56_56172, 1_56_56265,
1_57_57848, 1_60_60686, 1_62_62852, 1_63_63372, 1_63_63566, 1_65_65815, 1_65_65984, 1_66_66979,1_67_67033

their average intensity values to compute precision, recall and
F-measure metrics. The F-measure in this case is a measure of
accuracy of the saliency map to completely segment the whole
object from an image.
The proposed approach is compared with the nine features used as
individual saliency map generators, as described in the Section 3, and
seven state-of-the-art methods. The state-of-the-art methods selected
for comparison include Dense and Sparse Saliency (DSR) [56], Mani-
fold Ranking (MR) [57], Bayesian Saliency Model (BSM) [58], Cluster-
based Saliency (CS) [59], Low Rank Matrix Recovery (LRK) [60], Soft
Image Abstraction (SIA) [61] and Hierarchical Saliency (HS) [62]. The
benchmark methods are selected as they are recent, widely accepted
by the community and have made either their methods or their results
available for public use.
5. Results
This section shows the performance comparison of the ob-
tained results using the proposed approach with the individual
feature based methods and seven state-of-the-art methods.
that maintain more true positives at higher thresholds, resulting in
completely highlighted salient objects.
The high variability in terms of the PR curves of the features de-
monstrate the performance gaps amongst individual features. It is
noteworthy that the proposed XCSCA method is able to identify these
huge performance gaps and improve upon individual feature perfor-
mance by minimizing these gaps. The average PR results on all data-
sets in the bottom row in Fig. 4 demonstrate the performance gains of
38.8% and 2.7% achieved by the XCSCA method as compared with the
worst and best performing individual features, respectively.
5.1.2. Segmentation quality based comparison
Fig. 5 shows the segmentation performance of the individual fea-
tures in comparison with the proposed XCSCA method on all datasets.
The high variability of the precision, recall and F-measure metrics can
be observed from Fig. 5. The average F-measure results on all datasets
in Fig. 5(d) show performance improvements of 29.8% and 4.5% ob-
tained by the proposed XCSCA method as compared with the worst
and best performing individual features, respectively.
5.2. Comparison with state-of-the-art methods

5.1. Comparison with individual feature based methods
5.1.1. Precision recall curves based comparison
Fig. 4 shows the performance of individual features in comparison
with the proposed XCSCA method on all chosen benchmark datasets
in terms of PR curves (empirical as well as interpolated). The perfor-
mance of methods is summarized and ranked by the area under the
precision recall curve (AUCPR). It can be observed from Fig. 4 that at
0 threshold the precision values lie between 0.2 and 0.3 (at the right
extreme of the PR curves) for all datasets. This indicates that 20–30%
pixels belong to the annotated salient objects for all datasets. At the
other extreme of the PR curve for lower recall values, only the pro-
posed LCS method retains high precision values for most cases as
compared with the individual feature based methods. This behavior
corresponds to the smoother saliency maps of the proposed method
5.2.1. Precision recall curves based comparison
Fig. 6 shows the performance of seven state-of-the-art methods in
comparison with the proposed XCSCA method on all chosen bench-
mark datasets in terms of PR curves (empirical as well as interpolated).
In terms of the empirical PR curves, the proposed XCSCA method
outperforms all the state-of-the-art techniques on all datasets. The
overall superior quality of the saliency maps produced by the pro-
posed method can be observed from the average PR results. It can be
observed that several methods perform closer to the proposed method
on lower threshold values, however the proposed method maintains
the highest precision on higher threshold values (corresponding to
lower recall) as compared with the state-of-the-art methods. The in-
creased precision in the saliency results of the proposed method (at
similar recall values) can be attributed to the appropriate weighting of
features within each niche, resulting in suppression of background
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Fig. 9. Images grouped by three representative experienced rules from the population.
noise in cluttered scenarios.
In terms of the interpolated PR curves, DSR performs better than
all methods on the MSRA dataset and HS performs better than all
methods on the SOD dataset. However, on average XCSCA performs
better than all methods as shown in the bottom row in Fig. 6.
5.2.2. Segmentation quality based comparison
Fig. 7 shows the comparison of methods in terms of the quality
of their induced segmentation for MSRA, SOD, and SED2 datasets.
Fig. 7 suggests that none of the methods consistently outperforms
all other methods on this benchmark. However, on average the
proposed method outperforms all the state-of-the-art methods on
all datasets. It is noteworthy that the proposed method has con-
sistently the highest precision as compared with all other state-of-
the-art methods on all datasets, suggesting that the proposed
method maintains the highest number of true positives inside
object contours in the segmented output. The high precision and
F-measure performance of the proposed method can be ascribed
to the suppression of background noise achieved by the proposed
method especially on cluttered background images. Additionally,
the robust performance on this benchmark can be partly
117
associated to learning of adaptive thresholds during the training
stage of the proposed method.
6. Further discussions
6.1. Qualitative comparison
According to the precision recall curve based comparison, the
proposed method maintained higher precision than the state-of-
the-art methods on higher thresholds. Also the segmentation re-
sults reveal that the proposed method consistently obtained the
highest precision as compared with other methods. The robust-
ness of our approach in completely highlighting the salient objects
and uniform saliency assignment inside object contours as com-
pared with the state-of-the-art methods can be observed by the
visual comparison shown in Fig. 8. In most cases, the saliency
maps produced by the proposed LCS method exhibit the highest
quality of capturing the precise location of salient objects, com-
plete salient objects and effective suppression of background noise
as compared with the state-of-the-art methods.
118 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
Group 1
Fig. 10. Saliency grouping results corresponding to the images in Fig. 9. (For interpretation of the references to colors in this figure, the reader is referred to the web version
of this paper.)
A noteworthy quality of the proposed saliency maps is preservation
of object details as shown by the third and last row of Fig. 8, which
may benefit object recognition applications. The last row of Fig. 8
shows an example of saliency where the proposed method matches
the object details more precisely than depicted in the ground truth
and highlights both objects more evenly than state-of-the-art. The first
and fourth rows of Fig. 8 exhibit examples of proper highlighting of
complete salient object and suppression of background noise as
compared with the state-of-the-art techniques. The fourth row image
in Fig. 8 shows an example of a complex scene with an ambiguous
salient object. The state-of-the-art methods struggle in completely
highlighting the salient regions and include unwanted background
noise as in columns 5 and 6. The proposed method performs better
than the state-of-the-art in capturing the salient region, while sup-
pressing the unwanted background noise.
6.2. Analysis of evolved rules
To further analyze the niching scheme of the evolved solutions,
we performed an experiment on 500 test images from the MSRA
dataset. Three highly experienced representative rules were
Group 2
Group 3
selected from the population as shown in Table 2.
The images and corresponding saliency maps covered by each
rule were placed into distinct groups. To visualize image and sal-
iency groups formed by the rules, a dissimilarity matrix was cre-
ated by utilizing the feature distances for each image. Afterwards,
2D mapping and normalization of the data was performed to
display the images and saliency maps belonging to a group in the
form of an embedded image. As numerous images were covered
by each individual rule, it is not feasible to visualize all of them.
Hence only a fixed sized grid of size 200  200 is chosen and im-
age thumbnails of size 50  50 are used for visualization purposes.
The arrangement of images is subject to the dissimilarities be-
tween their feature distances in encoded feature space. Fig. 9
shows the image groups formed by the representative rules, while
Fig. 10 shows the corresponding saliency maps. The images are
placed into groups based on their feature composition and ar-
ranged in the embedded image according to their mutual dis-
tances in encoded feature space. The corresponding saliency maps
are computed by applying the corresponding classifier weights.
As the selected classifiers are highly general and experienced,
multiple images are covered by more than an individual classifier as
 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120
1 seven state-of-the-art combinatorial methods, in terms of precision–
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1 Average distance (Rule1)
Average distance (Rule2)
Average distance (Rule3)
0
f0 f1 f2 f3 f4 f5 f6 f7
Feature number
Fig. 11. Mean distance for images covered by an individual rule.
can be observed in Fig. 9. It is noteworthy that each group en-
compasses a variety of images belonging to different types. This is due
to the fact that images are grouped based on a variety of extracted
features where each feature captures a different property of the image.
The grouped saliency response in Fig. 10 elaborates the niching
property of the system as different characteristics saliency responses
are covered by different rules. In general group 2 includes saliency
outputs where high saliency is uniformly assigned inside object con-
tours as can be observed by the red highlighted cases. Groups 1 and
3 feature noisy saliency response where the background confusion
included in both cases differs in its characteristics. The red squares
show particularly clear examples for each group. It can be observed
that the background clutter included by group 1 saliency outputs
appears to capture valid segments of the image belonging to the
background. Conversely, the background noise included in group
3 saliency outputs is scattered or dispersed.
The saliency outputs enclosed in black boxes belong to images
covered by more than one rule. The different saliency response
generated by different rules can be easily observed for each du-
plicated image, e.g. the suppressed background saliency response
by the group 3 rule for the hand image as compared with the
saliency output in group 1 is noteworthy. Conversely, groups 2 and
3 generate similar saliency outputs for the cycle image.
Fig. 11 shows the average results for the feature distances of images
covered by an individual rule. This result depicts an important prop-
erty of the niching scheme of the proposed system. It can be observed
that the features f 4 − f 8 appear to be similar for all the images and
these distinct rules are created to accommodate the varying nature of
features f 0 − f 3 for these images. It is also noted that the influence of
f 4 and f 5 has been greatly reduced by the learnt weighting, i.e. ef-
fectively a “don't care”, showing the system can overcome the inclu-
sion of a limited number of redundant or irrelevant features.
7. Conclusions
The goal of this study was to investigate the effectiveness of
learning classifier systems in combining different image features to
detect salient objects from different complexity-level images. This goal
was successfully achieved by incorporating a novel encoding scheme
in a learning classifier system that computes actions using a linear
combination of features. The proposed approach effectively learned
different feature combinations for various types of images with dif-
ferent difficulty levels. The obtained results indicate that the proposed
approach outperforms nine individual feature based methods and
119
recall curves and F-measures. Further, it is observed that the proposed
method preserves more details of objects than the state-of-the-art
methods, which may benefit object recognition applications. Future
work includes the investigation of using more rich features (such as
computed in deep neural networks) and experimenting the LCS
technique on image classification benchmarks.
Acknowledgments
This work was supported in part by the Marsden Fund of the
New Zealand Government under contract VUW1209, admini-
strated by the Royal Society of New Zealand.
f8 References
[1] T. Liu, Z. Yuan, J. Sun, J. Wang, N. Zheng, X. Tang, H.-Y. Shum, Learning to detect
a salient object, IEEE Trans. Pattern Anal. Mach. Intell. 33 (2) (2011) 353–367.
[2] Y.-Z. Song, X. Bai, P.M. Hall, L. Wang, In search of perceptually salient group-
ings, IEEE Trans. Image Process. 20 (4) (2011) 935–947.
[3] S.S. Naqvi, W.N. Browne, C. Hollitt, Salient object detection via spectral mat-
ting, Pattern Recognit. 51 (2016) 209–224.
[4] R. Kimchi, M.A. Peterson, Figure-ground segmentation can occur without at-
tention, Psychol. Sci. 19 (7) (2008) 660–668.
[5] S. Liu, X. Bai, Discriminative features for image classification and retrieval,
Pattern Recognit. Lett. 33 (6) (2012) 744–751.
[6] H. Zhang, X. Bai, J. Zhou, J. Cheng, H. Zhao, Object detection via structural
feature selection and shape model, IEEE Trans. Image Process. 22 (12) (2013)
4984–4995.
[7] H. Jiang, J. Wang, Z. Yuan, Y. Wu, N. Zheng, S. Li, Salient object detection: a dis-
criminative regional feature integration approach, in: Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition, 2013, pp. 2083–2090.
[8] M.-M. Cheng, G.-X. Zhang, N.J. Mitra, X. Huang, S.-M. Hu, Global contrast based
salient region detection, in: Proceedings of IEEE Conference on Computer
Vision and Pattern Recognition, 2011, pp. 409–416.
[9] L. Shao, M. Brady, Specific object retrieval based on salient regions, Pattern
Recognit. 39 (10) (2006) 1932–1948.
[10] X. Yang, X. Qian, T. Mei, Learning salient visual word for scalable mobile image
retrieval, Pattern Recognit. 48 (10) (2015) 3093–3101.
[11] J. Wang, L. Quan, J. Sun, X. Tang, H.-Y. Shum, Picture Collage, in: Proceedings of IEEE
Conference on Computer Vision and Pattern Recognition, 2006, pp. 347–354.
[12] S. Goferman, A. Tal, L. Zelnik-Manor, Puzzle-like collage, Comput. Graph.
Forum 29 (2) (2010) 459–468.
[13] Y.-C. Chen, V.M. Patel, R. Chellappa, P.J. Phillips, Salient views and view-de-
pendent dictionaries for object recognition, Pattern Recognit. 48 (10) (2015)
3053–3066.
[14] L. Itti, Automatic foveation for video compression using a neurobiological model of
visual attention, IEEE Trans. Image Process. 13 (10) (2004) 1304–1318.
[15] D.A. Klein, S. Frintrop, Center-surround divergence of feature statistics for
salient object detection, in: Proceedings of International Conference on
Computer Vision, 2011, pp. 2214–2219.
[16] S. Goferman, L. Zelnik-Manor, A. Tal, Context-aware saliency detection, IEEE
Trans. Pattern Anal. Mach. Intell. 34 (10) (2012) 1915–1926.
[17] T. Judd, K. Ehinger, F. Durand, A. Torralba, Learning to predict where humans
look, in: Proceedings of International Conference on Computer Vision, 2009,
pp. 2106–2113.
[18] A. Borji, Boosting bottom-up and top-down visual features for saliency esti-
mation, in: Proceedings of IEEE Conference on Computer Vision and Pattern
Recognition, 2012, pp. 438–445.
[19] Q. Zhao, C. Koch, Learning a saliency map using fixated locations in natural
scenes, J. Vis. 11 (3) (2011) 1–15.
[20] Q. Zhao, C. Koch, Learning visual saliency by combining feature maps in a
nonlinear manner using adaboost, J. Vis. 12 (6) (2012) 1–15.
[21] J.H. Holland, L.B. Booker, M. Colombetti, M. Dorigo, D.E. Goldberg, S. Forrest, R.
L. Riolo, R.E. Smith, P.L. Lanzi, W. Stolzmann, S.W. Wilson, What is a learning
classifier system? in: Learning Classifier Systems, From Foundations to Ap-
plications, Springer, Berlin Heidelberg, 2000, pp. 3–32.
[22] L. Bull, T. Kovacs, Foundations of Learning Classifier Systems: An Introduction,
Springer, Berlin Heidelberg, 2005.
[23] M. Iqbal, S.S. Naqvi, W.N. Browne, C. Hollitt, M. Zhang, Salient Object Detection
Using Learning Classifier Systems that Compute Action Mappings, in: Proceedings
of the Genetic and Evolutionary Computation Conference, 2014, pp. 525–532.
[24] L. Bull, Applications of Learning Classifier Systems, Springer, Berlin Heidelberg,
2004.
[25] K. Shafi, T. Kovacs, H.A. Abbass, W. Zhu, Intrusion detection with evolutionary
learning classifier systems, Nat. Comput. 8 (1) (2009) 3–27.
[26] I. Kukenys, W.N. Browne, M. Zhang, Transparent, online image pattern clas-
sification using a learning classifier system, in: Applications of Evolutionary
120 M. Iqbal et al. / Pattern Recognition 60 (2016) 106–120

Computation, Lecture Notes in Computer Science, vol. 6624, Springer, Berlin,
Heidelberg, 2011, pp. 183–193.
[27] M. Behdad, L. Barone, T. French, M. Bennamoun, On XCSR for electronic fraud
detection, Evolut. Intell. 5 (2) (2012) 139–150.
[28] M. Iqbal, W.N. Browne, M. Zhang, XCSR with computed continuous action, in:
Proceedings of the Australasian Joint Conference on Artificial Intelligence,
2012, pp. 350–361.
[29] M. Iqbal, W.N. Browne, M. Zhang, Learning complex, overlapping and niche
imbalance boolean problems using XCS-based classifier systems, Evolut. Intell.
6 (2) (2013) 73–91.
[30] M. Iqbal, W.N. Browne, M. Zhang, Reusing building blocks of extracted
knowledge to solve complex, large-scale boolean problems, IEEE Trans. Evolut.
Comput. 18 (4) (2014) 465–480.
[31] M. Iqbal, W.N. Browne, M. Zhang, Extending XCS with cyclic graphs for scal-
ability on complex boolean problems, Evolut. Comput. (2015) http://dx.doi.
org/10.1162/EVCO_a_00167.
[32] S.W. Wilson, Classifier fitness based on accuracy, Evolut. Comput. 3 (2) (1995)
149–175.
[33] T. Kovacs, Evolving Optimal Populations with XCS Classifier Systems, Technical
Report CSR-96-17 and CSRP-9617, University of Birmingham, UK, 1996.
[34] S.W. Wilson, Generalization in the XCS Classifier System, in: Proceedings of
the Genetic Programming Conference, 1998, pp. 665–674.
[35] M.V. Butz, S.W. Wilson, An algorithmic description of XCS, Soft Comput. 6 (3–
4) (2002) 144–153.
[36] S.S. Naqvi, W.N. Browne, C. Hollitt, Optimizing visual attention models for
predicting human fixations using genetic algorithms, in: Proceedings of IEEE
Congress on Evolutionary Computation, 2013, pp. 1302–1309.
[37] L. Mai, F. Liu, Comparing salient object detection results without ground truth, in:
Proceedings of the European Conference on Computer Vision, 2014, pp. 76–91.
[38] N. Tong, H. Lu, Y. Zhang, X. Ruan, Salient object detection via global and local
cues, Pattern Recognit. 48 (10) (2015) 3258–3267.
[39] N. Singh, R. Arya, R. Agrawal, A. Novel, Approach to combine features for
salient object detection using constrained particle swarm optimization, Pat-
tern Recognit. 47 (4) (2014) 1731–1739.
[40] J. Yua, D. Taob, Y. Ruic, J. Cheng, Pairwise constraints based multiview features
fusion for scene classification, Pattern Recognit. 46 (2) (2013) 483–496.
[41] J. Yu, Y. Rui, D. Tao, Click prediction for web image reranking using multimodal
sparse coding, IEEE Trans. Image Process. 23 (5) (2014) 2019–2032.
[42] T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning: Data
Mining, Inference, and Prediction, Springer, New York, 2009.
[43] P.L. Lanzi, D. Loiacono, Classifier systems that compute action mappings, in:
Proceedings of the Genetic and Evolutionary Computation Conference, 2007,
pp. 1822–1829.
[44] T. Liu, Z. Yuan, J. Sun, J. Wang, N. Zheng, X. Tang, H.-Y. Shum, Learning to detect
a salient object, IEEE Trans. Pattern Anal. Mach. Intell. 33 (2) (2011) 353–367.
[45] H. Fu, X. Cao, Z. Tu, Cluster-based co-saliency detection, IEEE Trans. Image
Process. 22 (10) (2013) 3766–3778.
[46] B. Alexe, T. Deselaers, V. Ferrari, What is an Object? in: Proceedings of IEEE
Conference on Computer Vision and Pattern Recognition, 2010, pp. 73–80.
[47] S.S. Naqvi, W.N. Browne, C. Hollitt, Combining object-based local and global
feature statistics for salient object Search, in: Proceedings of International
Conference on Image and Vision Computing New Zealand, 2013, p. 6.
[48] X. Hou, L. Zhang, Saliency detection: a spectral residual approach, in: Pro-
ceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
2007, pp. 1–8.
[49] Y. Rubner, C. Tomasi, L.J. Guibas, The Earth Mover's distance as a metric for
image retrieval, Int. J. Comput. Vis. 40 (2) (2000) 99–121.
[50] O. Pele, M. Werman, Fast and robust Earth Mover's distances, in: Proceedings
of International Conference on Computer Vision, 2009, pp. 460–467.
[51] P.L. Lanzi, D. Loiacono, S.W. Wilson, D.E. Goldberg, Generalization in the XCSF
classifier system: analysis, improvement, and extension, Evolut. Comput. 15
(2) (2007) 133–168.
[52] A. Borji, D.N. Sihite, L. Itti, Salient object detection: a benchmark, in: Proceedings of
the European Conference on Computer Vision, Part II, 2012, pp. 414–429.
[53] S. Alpert, M. Galun, R. Basri, A. Brandt, Image segmentation by probabilistic
bottom-up aggregation and cue integration, in: Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition, 2007, pp. 1–8.
[54] R. Achanta, S. Hemami, F. Estrada, S. Süsstrunk, Frequency-tuned salient re-
gion detection, in: Proceedings of IEEE Conference on Computer Vision and
Pattern Recognition, 2009, pp. 1597–1604.
[55] K. Boyd, K.H. Eng, C.D. Page, Area under the precision–recall curve: point es-
timates and confidence intervals, in: Machine Learning and Knowledge Dis-
covery in Databases, 2013, pp. 451–466.
[56] X. Li, H. Lu, L. Zhang, X. Ruan, M.-H. Yang, Saliency detection via dense and
sparse reconstruction, in: IEEE International Conference on Computer Vision,
2013, pp. 2976–2983.
[57] C. Yang, L. Zhang, H. Lu, X. Ruan, M.-H. Yang, Saliency detection via graph-
based manifold ranking, in: Proceedings of the IEEE Conference on Computer
Vision and Pattern Recognition, 2013, pp. 3166–3173.
[58] Y. Xie, H. Lu, M.-H. Yang, Bayesian saliency via low and mid level cues, IEEE
Trans. Image Process. 22 (5) (2013) 1689–1698.
[59] H. Fu, X. Cao, Z. Tu, Cluster-based co-saliency detection, IEEE Trans. Image
Process. 22 (10) (2013) 3766–3778.
[60] X. Shen, Y. Wu, A unified approach to salient object detection via low rank
matrix recovery, in: Proceedings of the IEEE Conference on Computer Vision
and Pattern Recognition, 2012, pp. 853–860.
[61] M.-M. Cheng, J. Warrell, W.-Y. Lin, S. Zheng, V. Vineet, N. Crook, Efficient
salient region detection with soft image abstraction, in: IEEE International
Conference on Computer Vision, 2013, pp. 1529–1536.
[62] Q. Yan, L. Xu, J. Shi, J. Jia, Hierarchical saliency detection, in: Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition, 2013, pp. 1155–1162.

Muhammad Iqbal completed his PhD in Learning Classifier Systems at the School of Engineering and Computer Science, Victoria University of Wellington (VUW), New
Zealand. He is currently working as a postdoctoral research fellow at VUW, New Zealand. Iqbal's main research interests are in the area of evolutionary machine learning. His
research focuses on evolutionary image analysis and classification using transfer learning in genetic programming and learning classifier systems techniques. He is also
interested in medical image analysis, data mining, and scalability of evolutionary techniques.

Syed S. Naqvi has recently completed his PhD in Salient Object Detection at the School of Engineering and Computer Science, Victoria University of Wellington (VUW), New
Zealand. His research interests are in the area of visual saliency for generic object detection.

Will N. Browne is an Associate Professor at the School of Engineering and Computer Science, Victoria University of Wellington (VUW), New Zealand. His main area of
research is Applied Cognitive Systems. Essentially, how to use inspiration from natural intelligence to enable computers/machines/robots to behave usefully. This includes:
Cognitive Robotics, Learning Classifier Systems (a branch of evolutionary computation) and Modern Heuristics for industrial application.

Christopher Hollitt completed a BE(Hon) in Electrical and Electronic Engineering and a BSc(Hon) in Physics and Theoretical Physics at the University of Adelaide in 1994 and
1996 respectively. In 2007 he received a PhD from the same institution, having studied the control of high precision optomechanical systems for gravitational wave detection.
His current research work is in machine perception, image processing and robot control. The work spans a wide area, from fundamental problems in feature recognition,
through techniques for efficiently utilizing the limited sensory resources of a robot system, to high level applications of image processing.
Dr Hollitt is a senior lecturer in the School of Engineering and Computer Science at the Victoria University of Wellington.

Mengjie Zhang is a Professor of Computer Science at the School of Engineering and Computer Science, Victoria University of Wellington (VUW), New Zealand. His research is
mainly focused on evolutionary computation, particularly genetic programming, particle swarm optimization and learning classifier systems with application areas of image
analysis, multi-objective optimization, classification with unbalanced data, feature selection and reduction, and job shop scheduling. He has published over 400 academic
papers in refereed international journals and conferences. He has been serving as an associated editor or editorial board member for five international journals (including
IEEE Transactions on Evolutionary Computation and the Evolutionary Computation Journal) and as a reviewer of over fifteen international journals. He has been serving as a
steering committee member and a program committee member for over eighty international conferences.