You are on page 1of 7

1st.

Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ

An Investigate on Moving Object Tracking and


Detection in Images
2021 1st Babylon International Conference on Information Technology and Science (BICITS) | 978-1-6654-1573-6/21/$31.00 ©2021 IEEE | DOI: 10.1109/BICITS51482.2021.9509887

Abbas S. B. Sadkhan Dr. Seyed Reza Talebiyan Dr. Nazbanoo Farzaneh


Information Technology Dept. Computer Engineering Dept. Computer Engineering Dept.
Iraqi Media Network Faculty of Engineering, ImamReza Faculty of Engineering, ImamReza
Baghdad, Iraq International University International University
eng.abbas.sadkhan@ieee.org Mashhad, Iran Mashhad, Iran
talebiyan@gmail.com nazbanou.farzaneh@gmail.com

Abstract— Object detection and tracking one of the most the optimal tracking of moving objects for further analysis of
important areas of research due to routine change in motion of events. If there is only one moving object, the tracking
object and variation in scene size, occlusions, appearance problem is negligible. However, if there are several moving
variations, and ego-motion and illumination changes. objects that sometimes overlap or block each other, the
Specifically, feature selection is the vital role in object tracking. problem becomes more difficult to track. When multiple
Nowadays, moving objects tracking in real time is a new objects are removed after blocking, the correct location of
paradigm that has a great impact on video surveillance and each object must be found and matched so that it can
monitoring to detect and trace objects, and due to the constant
continuously identify and track multiple objects. [2].
change in the motion of the object and changing the size of the
scene, obstruction, changes in appearance and changes in In recent years, research on tracking semantic objects in
movement and brightness; it is an important area of research video has received much attention. Moving object tracking is
and a very important task in the field of security automation an important issue in video systems, such as surveillance
monitoring systems automatically. On the other hand, object systems, sports reporting, movie annotations, and traffic
tracking in a film is a matter of estimating positions and other management systems. However, the situation becomes more
relevant information about the movement of objects in the complicated if the context and moving objects change
film's visual sequences. The first step in such systems is to dynamically. In video analysis, the properties of moving
identify a moving object in the film. The second step is to track
objects such as color, texture and shape, etc. must be known
the detected object. The purpose of this paper is to analyze and
in order for the moving object to be recognizable and
review the previous approach towards object tracking, object
recognition and object detection in Images.
traceable [1-2].
Research into the tracking of semantic objects in video
Keywords—Object tracking, object recognition, object has attracted a great deal of attention in recent years due to
detection, Moving object tracking, video sequence. advances in camera miniaturization and cost reduction, and
I. INTRODUCTION has preferred it to large camera networks. This growing
number of cameras could lead to new signal processing
The modern world has a wealth of digital visual applications that use multiple sensors in large areas. Object
information. For the analysis and understanding of this tracking is a new way to detect moving objects beyond time
enormous collection of image data, there are several methods using the camera in video sequences. Their main purpose is
of image analysis. The object detection technique has a to relate the target objects, as well as the shape or features, to
significant application and provides a significant platform for the location of the objects in successive video sequences.
modern applications. The syntactic and semantic meaning of After that, sorting and recognizing objects is essential for
movies and images can be analyzed to retrieve information tracking objects in a computer vision program. In addition,
that is necessary and more. The possible utility of the image tracking is the first step in locating or detecting a moving
can be realized using this method. The most important object in the frame. Subsequently, the identified object can
content of an image is an object in it. For this reason, there is be divided into a trembling tree, birds, humans, and vehicles,
a significant need for better detection techniques [1]. and so on. However, in the image processing method, object
Object tracking involves following an object through a tracking using video sequences is a challenging task and it
sequence of frames. Tracking multiple moving objects to seems there are some issues like object-to-scene blockage,
monitor and investigate events is a challenging task, which is object-to-object, complex object motion, real-time
mostly important due to camera noise, light condition, processing requirements, as well as improper object shape in
blockage of objects and change of orientation of different this area. However, this tracking has many advantages, a few
objects. In recent years, video surveillance has been widely of which include traffic monitoring, robot vision,
used to control sensitive security areas, including highways, surveillance and security and video communications, public
borders, department stores, banks, and crowded public areas Therefore, the specific program requires optimal
places. Developments in computing power, high-speed business in terms of computing, communication and network
network infrastructure, and the availability of high-capacity accuracy. Revenue from computing and communications
storage devices have covered the way for low-cost, multi- depends on the amount and type of collaboration between
sensor video surveillance systems. In the meantime, keeping cameras to collect, distribute, and process data to validate
track on a moving object is an important task. Moving object decisions and reduce estimation and duplication errors [3].
tracking is based on the results of object detection to estimate

978-1-6654-1573-6/21/$31.00 ©2021 IEEE 69

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
Visual tracking, in general, begins with movement in
target recognition. In the first video frame, the target
manually or automatically detects and selects a person or
object within the frame via a face tracker. Then the
appearance of the object is modeled, which consists of two
components [3]:
• Visual display, which defines moving target features
with the help of visual features.
• Statistical modeling, which uses statistical learning
approaches to build a robust statistical model for
moving object recognition.
In view of the above, to conclude, in general, moving object
tracking faces the following three problems:

• Partitioning problem of moving object.


• Detecting a moving object.
• Moving target tracking in the obstruction.
In this survey we focus on various object detection and
tracking approaches. This paper is divided into following
sections: Section 2 describes the basics of the object
tracking. Section 3 consists of object detection, difficulties in
detection of objects and object detection techniques. Section
4 describes the object classification; Section 5 is literature
review on moving object tracking and detection and Section
6 outlines the conclusion.
Fig. 1. Main flow diagram of object tracking[5].
II. OBJECT TRACKING
Object tracking is an important task within the field of
computer vision. Tracking can be described as the context
for determining the orientation of an object over time as a III. OBJECT DETECTION
moving object in a scene. This is important in the field of In object detection, the object is identified from a video
computer vision due to the proliferation of powerful sequence and its pixels are clustered. There are various
computers and the growing need for automated monitoring techniques such as background subtraction, frame
systems, and it is widely used for applications such as differencing etc. for implementing object detection [4].
automatic monitoring, robotic monitoring, human-machine
interface, motion-based detection, vehicle, navigation, traffic A. Difficulties in Detection of Objects
monitoring and video indexing. A significant number of • Illumination: The Lightning in the same place may
these programs require reliable tracking methods that meet change during the day. Environmental conditions
real-time constraints and are challenging and complex in may also affect the brightness of an image [6].
terms of changes in object motion, scale and appearance,
• Positioning: The location of an object may vary
scene brightness and obstruction. Tracking results can be
from one position to the other, causing the system
affected by the lack of difference in one of the parameters.
not to recognize a particular object [6].
With regard to the above and other cases in tracking objects,
countless methods have been proposed [4]. In an object • Rotation: The system must be able to detect an
tracking program, the target object can be identified as image which is in rotated form, in case of methods
something that is attractive for analysis. In addition, the like shape based object detection [6].
moving object tracking is one of the main tasks in computer
vision and is widely applied in industrial vision, intelligent • Mirroring: An object detection system must be able
transportation systems and video surveillance. In recent to handle situations where mirrored images of the
years, the ability of machines to detect suspicious objects and objects are formed.
detect their activities in a particular environment has allowed • Occlusion: Occlusion is a condition where an object
an important part of the device to communicate with humans is not completely visible in the image. In such
in an effective and easy way. The current method for situations, there are chances of not detecting useful
analyzing and detecting a suspicious object usually requires portions [6].
exceptional markers attached to the suspicious object, which
prevents the widespread use of technology. • Scale: The scale or size of the object in a frame may
not be affecting a good object detection system.
In the object tracking, the video analysis involves,
moving object detection, object classification, frame to frame • Noisy Images: Image with deprived quality.
object tracking. Fig. 1 shows the main flow diagram of
object tracking [5].

70

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
B. Object Detection Techniques value, then the object is considered as a background. A fine
Object detection is the procedure of finding the non- background subtraction algorithm should be able to manage
stationary entities in the image sequences. Detection is the the objects which at first immersed in the background and
then foreground after a while. Moreover, a background
first step towards tracking the moving object in the video
subtraction algorithm should be computationally cost-
sequence. Fig. 2 shows a kind of Object detection [7].
effective and uses less memory, but be able to accurately
identify dynamic objects from a video sequence to adapt to
real-time applications [10].
IV. OBJECT CLASSIFICATION
In object classification, the objects are classified as
birds, humans, vehicles and other moving objects. Here, the
classification of objects is done based on their shape
features of the motion region [11]. The classification is as
follows:
Fig. 2. Object detection techniques [7]

TABLE I . METHODS OF OBJECT CLASSIFICATION [11].


1) Frame Differencing:
The existence of a dynamic object is found by
distinguishing two consecutive frames. The method uses a
reference background image for comparison purposes [8].
That is, simple separation followed by the threshold of
showing moving objects. This method works well for videos
with static background. There are advantages like good
accuracy and easiness in implementation. The demerits of
this method are difficulty in obtaining an absolute outline as
well as difficulty in performing with non-static background.
2) Optical Flow:
This method is to calculate the image optical flow field
and do a process called clustering based on the optical flow
distribution characteristics of image. Optical flow estimation
gives a two-dimensional vector field, which means, motion
field that denotes velocities and directions of each point of an
image sequence. Clustering is done based on the optical flow
distribution characteristics [8]. It defines direction and time
in a time sequence of two subsequent dimensional velocity
vectors. The advantage is that it can produce accurate object
motion information. The method is effective except for its A. Template Matching
time intense process. Optic flow can be used to calculate the
Background motion model that provides to even out an The method of finding image fractions that match the
image in the background. Individual motion can be detected pattern is called pattern matching, which is a simple process.
using the directional gradient of the image, which is not Pattern matching may be the best method for multiple
predicted by the background. It can even recognize motion in precise situations. This is true enough, but sometimes there is
a video from a non-stationary camera as well as dynamic a possibility that the discovered object is not new. The object
background but many of these techniques cannot be utilized can be identified using the pattern from which the video was
in real-time applications devoid of dedicated hardware. selected for a particular movie. Nevertheless, there is no
Moreover, this method is complex as it needs enormous optimal accuracy because all he knows is the best fit for each
number of calculations [9]. frame. The algorithm barely works if in case, the object is
constantly present in the video, or else it may show fake
3) Background Subtraction: detection [4].
This method is one of the most well-known and used
B. Color Based
motion detection methods. This technique distinguishes
between a target image and a background image to identify Object recognition using colors involved in objects is
the moving part and is usually efficient for presenting object also used extensively and provides a simple way to
information. i.e. the background frame is deducted from the implement it.
target frame. If the variation is larger than the threshold value This provides powerful information for object
(already set), then it is identified as the pixels of dynamic recognition. Color histograms are proven to be simple and
object, or else of the background. efficient and provide the edge. This information has been
The major step of this technique is the initialization and segmented into two approaches which is the part-based
updating of background frame. The correctness of result is approach and the efficient sub-window approach. Feature
affected by the effectiveness of these factors. It detects combination, photometric invariance and compactness are
dynamic systems using the pixel of the current image, by the three major features that need to be taken into account
pixel consisting of a background image (reference image) while integrating or appending the color attributes with the
during quantification. If the disparity is beyond the threshold object detection [4].

71

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
C. Shape Based Risha and Kumar [16] proposed a visual stream with
Recently, shape has been shown to be of great morphological function to detect the object in the film. More
importance in object recognition. They have been morphological operations are applied to achieve a clear
significantly examined to identify objects in appropriate real- moving target image. This study focused only on the static
world images. These features also acquire local features such camera. So must have to focus on the movement of the
as SIFT because most objects are represented and described camera and also recognize several objects in the video
by their shape and texture, such as different animals and frames.
other objects. They are most likely used to add an advantage Najva and Bijoy [17] combined the features of Tensor
to the local features [4]. with the SIFT approach to classify objects detected using the
Deep Neural Network (DNN), to provide a model for
V. LITERATURE REVIEW detecting and classifying objects in films. DNN is able to
Due to the above and other items in object tracking, manage the next big data with billions of parameters like the
many algorithms have been proposed in every parts are as human brain. The simulation results show that the proposed
follows for years 2013-2020: classification model offers more accurate results than the
existing methods, which is a combination of SIFT and tensile
Shen et al. [10] proposed a new hierarchical method for
properties for feature extraction and DNN for classification.
identifying a moving target based on temporal-spatial
salinity. they obtained refined diagnostic results using Weng et al. [18] proposed an algorithm to improve the
temporal and spatial information. Experimental results show performance of natural feature selection in the real world. In
that this method detects the moving objects in the film with addition, they used Powerful Power Acceleration (SURF) to
high accuracy and efficiency. In addition, this method does extract features from the image and identify the mobile live
not have a significant time delay effect compared to the HMI camera. These extracted features are calculated using the
method. However, this method evaluates the position of the gesture matrix through a homogeneous approach. The
object in all video frames as self-sufficient, false warnings simulation result shows that this algorithm easily and quickly
that are inevitable. detects and identifies the object from its natural features.
However, its speed and accuracy need to be improved.
Ben Ayed et al. [12] proposed a method for recognizing
text-based data in video frames by big data analysis. Video Zhang et al. [19] proposed a method combining frame
frames are parsed into several blocks of constant size and differences and non-parametric methods for tracking video
these blocks are analyzed using the HAR wavelet transform analysis. The simulation result proved that this approach
method. In addition, they used a neural network to classify performance is better than the traditional frame difference
textual and non-textual blocks. However, this study should and GMM. In addition, it can remove noise from the
focus on extracting areas in order to remove noisy areas as background, giving us the ability to move the object more
well as deleting text such as sections. accurately in applications such as food and crop tracking
analysis. However, this study, to ensure safety, should
Viswanath et al. [13] proposed and modeled an approach
increase the capability of the tracking system and provide an
using non-panoramic background modeling. Using this
intuitive supply chain for the subscriber.
method, they modeled the entire image element with a
spatio-temporal Gaussian. The simulation result shows that Houssineau et al. [20] propose a parameter-based
this method is able to detect moving materials with less false inequality space for unmodified camera networks, which
alarms. However, this method fails when not enough features extends to moving objects and integrates with sensor
are obtained from this section. calibration into a multi-Bayesian object tracking. The
performance of the framework obtained for camera
Soundrapandiyan and Mouli [14] have proposed a new
calibration is shown in real-time simulation, which leads to
and adaptive method for detecting pedestrians. In addition,
localization and tracking problems for a single object, as well
they separated the foreground objects from the background
as for multi-object tracking. In addition, they wanted to
due to the pixel intensity of the image. They then used a
extend the proposed method to other multi-object filters and
high-gain filter to enhance the front edges. The efficiency of
compare these methods for camera-based tracking as well as
the proposed method is evident from the results of subject
for camera calibration.
evaluation as well as objective evaluation with about 90% of
the pedestrian detection rate compared to other methods in Oiwa et al. [21] proposed a possible background model
the single image. In the future, they wanted to improve the for object tracking from video sequences. The simulation
performance of the method with higher detection speeds and results show the accuracy and effectiveness of this method
lower false positives, in line with sequence imaging methods. compared to the previous method above. However, this study
should focus on higher speeds as well as improved accuracy
Ramya and Rajeswari [15] have proposed a modified
in object tracking.
frame change method that uses the correlation between the
current image blocks and the background image to classify Markov Chain Monte Carlo (MCMC) method [22] is
the pixels as foreground and background. The blocks in the based on the use of probability-based estimates. MCMC is a
current image that are closely related to the background well-known type of Monte Carlo method that has been
image are considered as the background. For the other block, widely used in Bayesian inference and stochastic
a pixel comparison is performed to classify it as a foreground optimization. However, this method requires a large number
or background. Experiments have shown that this method of generated samples to be able to accurately calculate
improves the difference of frames, especially finding complex integrals and seek the optimal solution to an
accuracy. optimization problem.

72

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
Zhang et al. Have proposed a visual tracking framework search window. The image is then extracted in the search
based on particle swarm optimization (PSO) for tracking window. The graph, saturation color and HSV value of the
multiple objects [23]. Their study showed that the proposed extracted area are calculated for object modeling. The
tracking algorithm is able to track movement targets during Bhattacharyya distance between the HSV histograms of two
obstruction and maintain a correct identity. A new adaptive consecutive frames is then formulated as the MFPA target
discrete swarm optimization (ADSO) was proposed by Bae performance function, in which the maximum similarity of
et al. For real-time video tracking [24]. Experimental results both histograms is followed. Comparative experimental
show that ADSO performs better than traditional PSO, where results show that the developed algorithm has relatively good
faster tracking and control speeds as well as higher accuracy performance and accuracy.
are observed.
Moving tracking objects is an important yet challenging
Gao et al. Implemented a BAT-based tracker algorithm task. Many researchers use different methods in tracking
[25]. In their work, the sensitivity of tracking performance objects depending on the field of application. Prakash et al.
was analyzed according to BA parameters. Experimental [32] proposed a complex binary tree wavelet transform
results showed that BA with optimized parameters is able to algorithm for motion tracking objects in the video sequence.
track the moving target correctly and has even better Some researchers have used probabilistic methods such as
performance than PF, MS and PSO. the Kalman filter and the particle filter to track an object.
Chen et al. [33] Combined particle filters with averaging to
Fury et al. Showed that the Harmony Search Algorithm track video motion to improve the performance of particle
(HS) can be successfully used as a search strategy for visual filters.
tracking [26]. Their work showed that the HS-based
harmonic filter tracking system is more efficient in terms of Kalman filter and fuzzy logic have been found to have
speed and accuracy than its PF and noise-free KF. the corresponding superiority in various applications. Fuzzy
Subsequently, Gao et al. Have expanded the work of their theory has the ability to mimic human mental decision.
colleagues, and with the development of advanced HS, the Using fuzzy theory, intelligent systems can be adapted to
Improved Harmony Search (IHS), the world's best global their environment. This is why researchers are trying to use
search, compatible harmony search, and differential harmony fuzzy logic to adapt and overcome the shortcomings of the
search, these approaches have been applied to visual Kalman filter. In [34], the authors proposed a zero-phase
tracking. adaptive fuzzy Kalman filter to suppress the vibration of the
robotic surgical system.
A cuckoo search-based tracker has also been proposed in
[27]. The proposed tracking system works well in tracking Alqudaihi et al. [35] presented a hyper fuzzy extended
strong moving objects and has been shown to perform better Kalman filter for indoor security monitoring system. Li et al.
in terms of accuracy than PF, variable mean and PSO. [36] has worked on combining fuzzy adaptive strategy with
strong tracking scent unscented Kalman filter to solve the
The combination of PF with the modified Galaxy-based problem of misalignment angles in inertial navigation
search algorithm (MGbSA) has been studied in [28], so that system.
the optimal state search in the PF space is enhanced. It is
shown that the proposed method, due to its ability to manage Tseng et al. [37] has presented a sensor fusion method
nonlinear and non-Gaussian problems, can accurately combining cubature Kalman filter (CKF) and fuzzy logic
estimate the path of the target object. adaptive system (FLAS) for the integrated navigation
systems. All of the above mentioned works have applied
A new optimization algorithm, the Sinus Cosine fuzzy logic aided Kalman filter in different application areas.
Algorithm-Particle Swarm Optimization (SCA-PSO), is In our work, we have used Madmani based fuzzy logic to
proposed to solve the object tracking problem [29]. The adapt a linear Kalman filter which can solve an object
results showed that the proposed algorithm shows better tracking problem with improved performance.
performance compared to other existing methods, especially
MS, PF, PSO, BA, SCA and the combined gravity search TABLE II . SERVAY ON REFRENCE
algorithm (HGSA), according to the PSO exploitability and No Paper Title year Main Still open Methods
SCA exploration in the proposed SCA-PSO. Contribution problem used in
Research
To optimize object tracking, Kang et al. [30] proposed
the DeepHGSA Hybrid Gravity Search Algorithm [10] Moving object 2013 Detection a evaluates the A hierarchical
(DeepHGSA). In particular, the proposed DeepHGSA detection in moving target position of method based
aerial video the object in on temporal-
integrates the gravitational update component of the Gravity based on all video spatial
Search Algorithm (GSA), the cognitive and social spatiotemporal frames as salinity
saliency self-
components of the PSO, and the deep aggregation feature. sufficient
The test results show that the hybrid algorithm improves and false
ADSO accuracy and resistance by 50.6% and 56.9%, warnings
respectively, due to its fast PSO convergence feature and
GSA capability to take full advantage of all current [12] Map Reduce- 2015 Detection focus on Video frames
based text text-based extracting parsed into
information. detection in big objects in areas in several blocks
data natural video order to of constant
In [31], the authors proposed a new moving object scene videos remove size and
tracking algorithm based on the Flower Pollination noisy areas analyzed
using the
Modification (MFPA) algorithm. To search for the position HAR wavelet
of a moving object, a search window is used with the transform
properties of the center coordinates and the width of the

73

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
method bat algorithm Tracking speeds as algorithm
well as with
higher optimized
[13] Background 2015 non- fails when Using this
accuracy parameters
Modelling panoramic not enough method,
from a Moving features are modeled the
Camera obtained entire image [29] Object A new
modeling
element with Tracking optimization
a spatial- algorithm, the
temporal Sinus Cosine
Gaussian Algorithm-
Particle
Swarm
[14] Adaptive 2015 Pedestrian improve the used a high-
Optimization
Pedestrian Detection performance gain filter to
(SCA-PSO)
Detection in of the enhance the
Infrared method with front edges
Images Using higher ]30[ A hybrid 2018 Object Fast A Hybrid
Background detection gravitational Tracking convergence gravitational
Subtraction speeds and search and increase search with a
and Local lower false algorithm with accuracy and
deep
Thresholding positives swarm resistance
intelligence convolution
and deep feature
[17] SIFT and 2016 Detection offers more combined the (DeepHGSA)
convolutional
Tensor-Based objects accurate features of
feature for integrates the
Object Tensor with
object tracking gravitational
Detection and the SIFT
optimization update
Classification approach to
in Videos classify component of
Using Deep objects the
Neural detected Gravitational
Networks using the Search
Deep Neural
Algorithm
Network
(DNN) (GSA), the
cognitive and
social
[19] Moving Object 2016 Tracking to ensure combining
Detection video analysis safety, frame components
Based on should differences of the PSO,
Nonparametric increase the and non- and the deep
Methods and capability of parametric convolution
Frame the tracking methods for feature
Difference for system tracking
Traceability video analysis
Video Analysis ]31[ Effective 2019 Object Has A search
moving object Tracking efficiency window with
tracking using and center
[20] A Unified 2016 Object extend to a parameter- modified accuracy,
Approach for Tracking other multi- based coordinate
flower relatively
Multi-Object object filters inequality pollination good properties and
Triangulation, and compare space for algorithm for performance a search
Tracking and unmodified visible image window
Camera camera sequences width is used
Calibration. networks, under to determine
IEEE Trans which complicated
extends to the position
background
moving of a moving
objects object. The
image is then
[22] Orthogonal 2016 Requires a Based on the extracted in
parallel large use of the search
MCMC number of probability- window.
methods for generated based
sampling and samples to estimates and
optimization be able to the Monte
calculate Carlo method
complex
integrals
VI. CONCLUSION
accurately This paper discusses tracking and tracing techniques that
can enable readers to quickly absorb techniques. The process
[24] A novel real 2016 Object track proposed a
time video Tracking movement visual
of tracking an object can be divided into three groups: object
tracking targets tracking detection, object classification, and object tracking. This
framework during framework article describes different methods for identifying moving
using adaptive obstruction based on new
discrete swarm and maintain adaptive
objects as well as tracking them. Various techniques such as
optimization a correct discrete background subtraction, optical flux, and frame separation
identity swarm are illustrated, along with commonly used object tracking
optimization
(ADSO) for models. Finally, some sources of literature on tracking and
tracking tracing moving objects were mentioned.
multiple
objects Due to the extent of this issue, more studies are needed in
this field.
[25] A novel visual 2016 Object Faster Implemented
tracking tracking and a BAT-based
method using handling tracker

74

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.
1st. Babylon International Conference on Information Technology and Science 2021 (BICITS 2021)- Babil- IRAQ
REFERENCES [21] D. Oiwa, S. Fukui, Iwahori, Y., Nakamura, T. & Bhuyan, M.K., 2016.
Tracking with Probabilistic Background Model by Density Forests,
[1] M. Tiwari, & R. Singhai, (2017). A review of detection and tracking in: Proc. of IEEE/ACIS. IEEE, Okayama, Japan, pp. 391–396.
of object from image and video sequences. Int. J. Comput. Intell. Res,
13(5), 745-765. [22] L. Martino, V. Elvira, D. Luengo, J. Corander, F. Louzada,
Orthogonal parallel MCMC methods for sampling and optimization,
[2] X. Chen, X. Wang, & J. Xuan, (2018). Tracking multiple moving Digit. Signal Process. 58 (2016) 64–84.
objects using unscented Kalman filtering techniques. arXiv preprint
arXiv:1802.01235. [23] X. Zhang, W. Hu, W. Qu, S. Maybank, Multiple object tracking via
speciesbased particle swarm optimization, IEEE Trans. Circuits Syst.
[3] K.M.Ong, K. M., Ong, P., Sia, C. K., & Low, E. S. (2019). Effective Video Technol. 20 (11) (2010) 1590–1602
moving object tracking using modified flower pollination algorithm
for visible image sequences under complicated background. Applied [24] C. Bae, K. Kang, G. Liu, Y.Y. Chung, A novel real time video
Soft Computing, 83, 105625. tracking framework using adaptive discrete swarm optimization,
Expert Syst. Appl. 64 (2016) 385–399.
[4] M. Yazdi, & T. Bouwmans. (2018). New trends on moving object
[25] M.-L. Gao, et al., A novel visual tracking method using bat algorithm,
detection in video images captured by a moving camera: A
Neurocomputing 177 (2016) 612–619.
survey. Computer Science Review, 28, 157-177.
[5] S.R. Balaji, & S. Karthikeyan, (2017, January). A survey on moving [26] J. Fourie, S. Mills, R. Green, Harmony filter: A robust visual tracking
object tracking using image processing. In 2017 11th international system using the improved harmony search algorithm, Image Vis.
conference on intelligent systems and control (ISCO) (pp. 469-474). Comput. 28 (12) (2010) 1702–1716.
IEEE. [27] M.-L. Gao, L.-J. Yin, G.-F. Zou, H.-T. Li, W. Liu, Visual tracking
[6] P. Panchal, G. Prajapati, S. Patel, H. Shah and J. Nasriwala, “A method based on cuckoo search algorithm, Opt. Eng. 54 (7) (2015)
Review on Object Detection and Tracking Methods”, International 073105
Journal of Research in Emerging Science and Technology”, Volume- [28] F. Sardari, M. Ebrahimi Moghaddam, A hybrid occlusion free object
2, Issue-1, January-2015. tracking method using particle filter and modified galaxy based
[7] J. Zhu, Z. Wang, S. Wang, & S. Chen, (2020). Moving Object search meta-heuristic algorithm, Appl. Soft Comput. J. 50 (2017)
Detection Based on Background Compensation and Deep Learning. 280–299.
Symmetry, 12(12), 1965. [29] H. Nenavath, D.R. Kumar Jatoth, D.S. Das, A synergy of the sine-
[8] H. L. Prathap, M. R. Sunitha & P. Scholar, (2016). A Survey on cosine algorithm and particle swarm optimizer for improved global
Moving Object Detection and Tracking Techniques. International optimization and object tracking, Swarm Evol. Comput. (August
Journal of Engineering and Computer Science. 2017) (2018) 1–30.
[9] R. Hemangi Patil, K. Prof. K. ,S. Bhagat,” Detection and Tracking of [30] K. Kang, C. Bae, H.W.F. Yeung, Y.Y. Chung, A hybrid gravitational
Moving Object: A Survey”, Int. Journal of Engineering Research and search algorithm with swarm intelligence and deep convolutional
Applications, Vol. 5, Issue 11, (Part - 5) November 2015. feature for object tracking optimization, Appl. Soft Comput. J. 66
(2018) 319–329.
[10] H. Shen, Li, S., Zhu, C.,H. Chang, & J. Zhang, 2013. Moving object
[31] Ong, K. M., Ong, P., Sia, C. K., & Low, E. S. (2019). Effective
detection in aerial video based on spatiotemporal saliency. Chinese J.
moving object tracking using modified flower pollination algorithm
Aeronaut. 26, 1211–1217. doi: 10.1016/j.cja.2013.07.038.
for visible image sequences under complicated background. Applied
[11] J. Rupa, S.Rakibe, B. D.Patil, "Background Subtraction Algorithm Soft Computing, 83, 105625.
Based Human Motion Detection", Intelllational Journal of Scientific
and Research Publications, May 2013. [32] O. Prakash, M. Khare, C. M. Sharma, and A. S. K. Kushwaha,
“Moving object tracking in video sequences based on energy of
[12] Ben Ayed, A., Ben Halima, M., & Alimi, A.M., 2015. MapReduce- Daubechies complex wavelet transform,” in Proceedings of National
based text detection in big data natural scene videos. Procedia Conference on Communication Technologies & Its Impact on Next
Comput. Sci. 53, 216– 223. doi: 10.1016/j.procs.2015.07.297 Generation Computing (CTNGC), Mohan Nagar, India, 2012, pp. 6-
[13] A. Viswanath, R. Kumari, & V. Senthamilarasu, 2015. Background 10.
Modelling from a Moving Camera. Procedia - Procedia Comput. Sci. [33] K. Chen, D. Li, Q. Huang, and L. E. Banta, “Video motion tracking
58, 289–296. doi: 10.1016/j.procs.2015.08.023 using enhanced particle filtering with Mean-shift,” in Proceedings of
[14] R. Soundrapandiyan, & Mouli, P.V.S.S.R.C., 2015. Adaptive the 3rd International Congress on Image and Signal Processing,
Pedestrian Detection in Infrared Images Using Background Yantai, China, 2010, pp. 387-391.
Subtraction and Local Thresholding. Procedia Comput. Sci. 58, 706– https://doi.org/10.1109/CISP.2010.5648016.
713. doi: 10.1016/j.procs.2015.08.091 [34] H. Sang, C. Yang, Fen. Liu, J. Yun, G. Jin, and F. Chen, “A zero
[15] P. Ramya, & R. Rajeswari, 2016. A Modified Frame Difference phase adaptive fuzzy Kalman filter for physiological tremor
Method Using Correlation Coefficient for Background Subtraction. suppression in robotically assisted minimally invasive surgery,” The
Procedia Comput. Sci. 93, 478–485. doi: 10.1016/j.procs.2016.07.236 International Journal of Medical Robotics and Computer Assisted
[16] K.P. Risha, & A.C. Kumar, 2016. Novel Method of Detecting Surgery, vol. 12, no. 4, pp. 658-669, 2016.
Moving Object in Video. Procedia Technol. 24, 1055–1060. doi: https://doi.org/10.1002/rcs.1741.
10.1016/j.protcy.2016.05.235 [35] A. Alqudaihi, M. Johdy, and H. Abdel-Aty-Johdy, “Applying hyper-
[17] N. Najva, & K.E. Bijoy, 2016. SIFT and Tensor-Based Object fuzzy extended Kalman filter to indoor security monitoring,”
Detection and Classification in Videos Using Deep Neural Networks. Automatic Control and System Engineering, vol. 17, no. 1, pp. 41-51,
Procedia Comput. Sci. 93, 351–358. doi: 10.1016/j.procs.2016.07.220 2017.
[18] E.N.G. Weng, R.U. Khan, S.A.Z. Adruce, & Bee, O.Y., 2013. [36] J. Li, N. Song, G. Yang, and R. Jiang, “Fuzzy adaptive strong
Objects Tracking from Natural Features in Mobile Augmented tracking scaled unscented Kalman filter for initial alignment of large
Reality. Procedia - Soc. Behav. Sci. 97, 753–760. doi: misalignment angles,” Review of Scientific Instruments, vol. 87, no.
10.1016/j.sbspro.2013.10.297 7, article no. 075118, 2016. https://doi.org/10.1063/1.4959561.
[19] J. Zhang, J. Cao, & B. Mao, 2016. Moving Object Detection Based [37] C. H. Tseng, S. F. Lin, and D. J. Jwo, “Fuzzy adaptive cubature
on Nonparametric Methods and Frame Difference for Traceability Kalman filter for integrated navigation system,” Sensors, vol. 16, no.
Video Analysis. Procedia Comput. Sci. 91, 995–1000. doi: 8, article no. 1167, 2016. https://doi. org/10.3390/s16081167.
10.1016/j.procs.2016.07.132
[20] J. Houssineau, D.E.Clark., S. Ivekovic, Lee, C.S. & Franco, J., 2016.
A Unified Approach for Multi-Object Triangulation, Tracking and
Camera Calibration. IEEE Trans. Signal Process. 64, 2934–2948.
doi:10.1109/TSP.2016.2523454

75

Authorized licensed use limited to: ULAKBIM UASL - Uluslararasi Kibris Universitesi. Downloaded on March 22,2023 at 05:03:04 UTC from IEEE Xplore. Restrictions apply.

You might also like