Survey on Segmentation of Partially Overlapping Objects
0 likes34 views
This document summarizes several existing methods for segmenting partially overlapping objects in digital images. It discusses challenges in segmenting overlapping objects and different approaches researchers have used, including watershed-based methods, graph cuts algorithms, active shape models, and level set methods. The goal of segmentation is to partition an image into meaningful regions to analyze objects and boundaries. Efficient segmentation of overlapping objects remains an important challenge in image processing.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 796
Survey on Segmentation of Partially Overlapping Objects
Dyni Thomas1, Joe Mathew Jacob2
1PG student, Dept. of Computer Engineering, VJCET, Vazhakulam, Kerala, India
2Assistant Professor, Dept. of Computer Engineering, VJCET, Vazhakulam, Kerala, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Digital image segmentation is the process of
partitioning an image into multiple segments or sets of pixels,
also known as super-pixels. The goal of segmentation is to
simplify change the representation of a digital image into
something that is more meaningful and easier to analyze. It is
typically used to locate objects and boundaries such as lines
and curves to detect the relevant information in images.
Efficient image segmentation is one of the most critical tasks
in automatic image processing. Hence this literature survey
studies various existing methods presents a segmentation of
clustered partially overlapping objects.
Key Words: Digital image segmentation, Overlapping
objects, Cell nuclei, Histopathology.
1. INTRODUCTION
Segmentation of overlapping objectsisanincreasingamount
of various computer applications such as Biomedical
industry. Image segmentation is defined as a process of
partitioning a digital image into homogenous groups such
that each region is homogenous but the union of two
adjacent regions is not homogenous. Also the image
segmentation has been interpreted differently for different
applications in image processing.
Image segmentation is a difficult area of research for image
processing field. But it is still a challenging task for
researchers and developerstodevelopa universal technique
for image segmentation techniques. Digital image
segmentation is used to differentiate different objects in the
image. Since the image is divided into foreground and
background, whereas foreground of image is related to the
region of interest part in image and background istherest of
the image. Hence, image segmentation will separate these
two parts from one another. In this paper different method
for image segmentation in partially overlapping objects are
studied.
2. LITERATURE SURVEY
In[1] Chanho Jung and Changick Kim proposed a novel
watershed-based method for segmentation biomedical
images. First apply the distance transform to the clustered
nuclei. A marker extraction scheme based on the H-minima
transform is introduced to obtain the feasible segmentation
result from the distance map. The H-minima or H-maxima
transform is a powerful mathematical tool to suppress
undesired minima or maxima. Performing the H-minima
transform on the inverse distance image can effectively
decrease oversegmentation.
In[2] Yousef Al-Kofahi, Wiem Lassoued proposed an
automatic segmentation, that is an essential step in
biomedical image analysis. For progress analysis, there is a
need to improve accuracy, speed, level of automation, and
adaptability to new applications. This work presents a
robust method for segmenting the cell nuclei. The image
foreground is extracted using a graph-cuts-based
binarization. Next, nuclear seed points are detected by a
method, that combining multiscale Laplacian-of-Gaussian
filtering. The multiscale LoG filtering is constrainedbyusing
a distance-map-based adaptive scale selection process.
These points are used to perform an initial segmentation
that is refined using a second graph-cuts-based algorithm.
The graph based algorithm method combining the alpha
expansions and graph coloring to reduce computational
complexity.
In[3] Yan Nei Law, Hwee Kuan Lee proposed a variant of the
Mumford-Shah model for the segmentation of a pair of
overlapping objects with additive intensity value. The basic
idea of using Mumford-Shah model is, it can handle
gracefully complex situations and is very robust to noise. It
does not only determine distinct objects in the image, but
also recover the possibly multiple membership of the pixels.
To accomplish this feature, some apriori knowledge about
the smoothness of the object boundaryisintegratedintothis
model. Additivity is imposed through a softconstraintwhich
allows the user to control the degree of additivity, and it is
more accurate than the hard constraint. The main
contribution in this work is to construct a relaxed additive
model, called the soft additive model, which can overcome
the problems of the hard additive model. Another
contribution is to provide a stability analysis of the soft
model to demonstrate the robustness.
In[4] Marina E. Plissiti and Christophoros proposed a
segmentation methodofoverlapping nuclei,whichcombines
the segmentation of an image with Active Shape Model and
the representation of an object using modal analysis.Thus,a
physical model is adopted in the training phase, in whichthe
parameters are the variations of themodesofthe model.The](https://image.slidesharecdn.com/irjet-v4i2155-171117093837/85/Survey-on-Segmentation-of-Partially-Overlapping-Objects-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 797
attributes of objects in image such as, nuclei shapes are
expressed in terms of modal analysis and in the training
phase the modal distribution is estimated.Therefore,a more
compact description of the shape model is obtained. Next,
develop a framework for the deformation of an active
physical model similar to ASM, for the detection of an
unknown new nucleus in images containing two overlapped
nuclei.
In[5] Fuyong Xing, David J. Foran proposed an biomedical
image analysis of histopathology specimens that, provide
support for early detection of breast cancer. Automated
segmentation of the cells comprising imaged tissue
microarrays (TMAs) is a prerequisite for quantitative
analysis. The segmentation algorithm called touching cell
segmentation is composed of two steps. It begins with a fast,
accurate, reliable object center localization approach that
utilizes single path voting followed by mean-shiftclustering.
Mean shift clustering provides accurate seed detection and
robust touching object localization. Next,thecontourofeach
cell is obtained by using a level set algorithm based on an
interactive model.
In[6] Chiwoo Park, Member, IEEE, Jianhua Z. Huang
proposed a method that enables automated morphology
analysis of partially overlapping nanoparticles in electron
micrographs. The method adopts a two-stage approach:The
first stage executes the task of particle separation, that is, a
modified ultimate erosion process is developed for
decomposing a mixture of particles. The second stage
simultaneously conducts the tasks of contour inference and
the shape classification by using a set of evidences becomes
inputs to a Gaussian mixture model.
In[7] Sahar Zafari, Tuomas Eerola proposed a segmentation
of partially overlapping objects with a known shapethatcan
be approximated using an ellipse. The method using the
silhouette images requires only that the foreground and
background objects can be distinguished from each other.
The work makes two contributions; the first contribution of
this work is the combined method of Bounded Erosion and
Fast Radial Symmetry (BE-FRS) for seedpoint extraction
from a group of partially overlapping objects. The second
contribution is integration of the BE-FRS method into the
segmentation of partially overlapping objects, that enabling
some improvements compared to existing methods.
3. CONCLUSIONS
Different methods to reduce the variations such as pose,
shape and size of partially overlapping objects are studied.
Each method has different ways to analyze the overlapping
objects. But the Influence of various variations is not yet
eliminated. From this study recognized that the basic
information is needed to analyze the overlapping objects
such as, shapes. To overcome these, considered as an
efficient method to segment multiple partially overlapping
objects and detects its original shapes. From this efficient
method we have to easily analyze the overlapping objects.
REFERENCES
[1] Chanho Jung and Changick Kim,“ Segmenting Clustered
Nuclei Using H- minima Transform-Based Marker
Extraction and Contour Parameterization, "IEEE
Transactions on Biomedical Engineering,Vol.57,No.10,
October 2010.
[2] Yousef Al-Kofahi, Wiem Lassoued, William Lee, and
Badrinath Roysam,“ Improved Automatic Detection and
Segmentation of Cell Nuclei in Histopathology Images,
"IEEE Transactions on Biomedical Engineering, Vol. 57,
No. 4, April 2010
[3] Yan Nei Law, Hwee Kuan Lee, Chaoqiang Liu, and Andy
M. Yip, “ A Variational Model for Segmentation of
Overlapping Objects With Additive Intensity Value, "IEEE
Transactions on Image Processing, Vol. 20, No. 6, June
2011.
[4] Marina E. Plissiti and Christophoros Nikou, “
Overlapping Cell Nuclei Segmentation Using a Spatially
Adaptive Active Physical Model, "IEEE Transactions on
Image Processing, Vol. 21, No. 11, November 2012.
[5] Xin Qi, Fuyong Xing, David J. Foran,andLinYang,“Robust
Segmentation of Overlapping Cells in Histopathology
Specimens Using Parallel Seed Detection and Repulsive
Level Set, "IEEE Transactions on Biomedical
Engineering, Vol. 59, No. 3, March 2012.
[6] Chiwoo Park, Jianhua Z. Huang, Jim X. Ji, “ Inference, and
Classification of Partially Overlapping Nanoparticles,
"IEEE Transactions on Pattern Analysis and Machine
Intelligence, Vol. 35, No. 3, March 2013.
[7] Sahar Zafari, Tuomas Eerola, Jouni Sampo, “
Segmentation of Overlapping Elliptical Objects in
Silhouette Images, "IEEE Transactions on Image
Processing, Vol. 24, No. 12, December 2015.](https://image.slidesharecdn.com/irjet-v4i2155-171117093837/85/Survey-on-Segmentation-of-Partially-Overlapping-Objects-2-320.jpg)
Survey on Segmentation of Partially Overlapping Objects
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056 Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 796 Survey on Segmentation of Partially Overlapping Objects Dyni Thomas1, Joe Mathew Jacob2 1PG student, Dept. of Computer Engineering, VJCET, Vazhakulam, Kerala, India 2Assistant Professor, Dept. of Computer Engineering, VJCET, Vazhakulam, Kerala, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - Digital image segmentation is the process of partitioning an image into multiple segments or sets of pixels, also known as super-pixels. The goal of segmentation is to simplify change the representation of a digital image into something that is more meaningful and easier to analyze. It is typically used to locate objects and boundaries such as lines and curves to detect the relevant information in images. Efficient image segmentation is one of the most critical tasks in automatic image processing. Hence this literature survey studies various existing methods presents a segmentation of clustered partially overlapping objects. Key Words: Digital image segmentation, Overlapping objects, Cell nuclei, Histopathology. 1. INTRODUCTION Segmentation of overlapping objectsisanincreasingamount of various computer applications such as Biomedical industry. Image segmentation is defined as a process of partitioning a digital image into homogenous groups such that each region is homogenous but the union of two adjacent regions is not homogenous. Also the image segmentation has been interpreted differently for different applications in image processing. Image segmentation is a difficult area of research for image processing field. But it is still a challenging task for researchers and developerstodevelopa universal technique for image segmentation techniques. Digital image segmentation is used to differentiate different objects in the image. Since the image is divided into foreground and background, whereas foreground of image is related to the region of interest part in image and background istherest of the image. Hence, image segmentation will separate these two parts from one another. In this paper different method for image segmentation in partially overlapping objects are studied. 2. LITERATURE SURVEY In[1] Chanho Jung and Changick Kim proposed a novel watershed-based method for segmentation biomedical images. First apply the distance transform to the clustered nuclei. A marker extraction scheme based on the H-minima transform is introduced to obtain the feasible segmentation result from the distance map. The H-minima or H-maxima transform is a powerful mathematical tool to suppress undesired minima or maxima. Performing the H-minima transform on the inverse distance image can effectively decrease oversegmentation. In[2] Yousef Al-Kofahi, Wiem Lassoued proposed an automatic segmentation, that is an essential step in biomedical image analysis. For progress analysis, there is a need to improve accuracy, speed, level of automation, and adaptability to new applications. This work presents a robust method for segmenting the cell nuclei. The image foreground is extracted using a graph-cuts-based binarization. Next, nuclear seed points are detected by a method, that combining multiscale Laplacian-of-Gaussian filtering. The multiscale LoG filtering is constrainedbyusing a distance-map-based adaptive scale selection process. These points are used to perform an initial segmentation that is refined using a second graph-cuts-based algorithm. The graph based algorithm method combining the alpha expansions and graph coloring to reduce computational complexity. In[3] Yan Nei Law, Hwee Kuan Lee proposed a variant of the Mumford-Shah model for the segmentation of a pair of overlapping objects with additive intensity value. The basic idea of using Mumford-Shah model is, it can handle gracefully complex situations and is very robust to noise. It does not only determine distinct objects in the image, but also recover the possibly multiple membership of the pixels. To accomplish this feature, some apriori knowledge about the smoothness of the object boundaryisintegratedintothis model. Additivity is imposed through a softconstraintwhich allows the user to control the degree of additivity, and it is more accurate than the hard constraint. The main contribution in this work is to construct a relaxed additive model, called the soft additive model, which can overcome the problems of the hard additive model. Another contribution is to provide a stability analysis of the soft model to demonstrate the robustness. In[4] Marina E. Plissiti and Christophoros proposed a segmentation methodofoverlapping nuclei,whichcombines the segmentation of an image with Active Shape Model and the representation of an object using modal analysis.Thus,a physical model is adopted in the training phase, in whichthe parameters are the variations of themodesofthe model.The
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056 Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 797 attributes of objects in image such as, nuclei shapes are expressed in terms of modal analysis and in the training phase the modal distribution is estimated.Therefore,a more compact description of the shape model is obtained. Next, develop a framework for the deformation of an active physical model similar to ASM, for the detection of an unknown new nucleus in images containing two overlapped nuclei. In[5] Fuyong Xing, David J. Foran proposed an biomedical image analysis of histopathology specimens that, provide support for early detection of breast cancer. Automated segmentation of the cells comprising imaged tissue microarrays (TMAs) is a prerequisite for quantitative analysis. The segmentation algorithm called touching cell segmentation is composed of two steps. It begins with a fast, accurate, reliable object center localization approach that utilizes single path voting followed by mean-shiftclustering. Mean shift clustering provides accurate seed detection and robust touching object localization. Next,thecontourofeach cell is obtained by using a level set algorithm based on an interactive model. In[6] Chiwoo Park, Member, IEEE, Jianhua Z. Huang proposed a method that enables automated morphology analysis of partially overlapping nanoparticles in electron micrographs. The method adopts a two-stage approach:The first stage executes the task of particle separation, that is, a modified ultimate erosion process is developed for decomposing a mixture of particles. The second stage simultaneously conducts the tasks of contour inference and the shape classification by using a set of evidences becomes inputs to a Gaussian mixture model. In[7] Sahar Zafari, Tuomas Eerola proposed a segmentation of partially overlapping objects with a known shapethatcan be approximated using an ellipse. The method using the silhouette images requires only that the foreground and background objects can be distinguished from each other. The work makes two contributions; the first contribution of this work is the combined method of Bounded Erosion and Fast Radial Symmetry (BE-FRS) for seedpoint extraction from a group of partially overlapping objects. The second contribution is integration of the BE-FRS method into the segmentation of partially overlapping objects, that enabling some improvements compared to existing methods. 3. CONCLUSIONS Different methods to reduce the variations such as pose, shape and size of partially overlapping objects are studied. Each method has different ways to analyze the overlapping objects. But the Influence of various variations is not yet eliminated. From this study recognized that the basic information is needed to analyze the overlapping objects such as, shapes. To overcome these, considered as an efficient method to segment multiple partially overlapping objects and detects its original shapes. From this efficient method we have to easily analyze the overlapping objects. REFERENCES [1] Chanho Jung and Changick Kim,“ Segmenting Clustered Nuclei Using H- minima Transform-Based Marker Extraction and Contour Parameterization, "IEEE Transactions on Biomedical Engineering,Vol.57,No.10, October 2010. [2] Yousef Al-Kofahi, Wiem Lassoued, William Lee, and Badrinath Roysam,“ Improved Automatic Detection and Segmentation of Cell Nuclei in Histopathology Images, "IEEE Transactions on Biomedical Engineering, Vol. 57, No. 4, April 2010 [3] Yan Nei Law, Hwee Kuan Lee, Chaoqiang Liu, and Andy M. Yip, “ A Variational Model for Segmentation of Overlapping Objects With Additive Intensity Value, "IEEE Transactions on Image Processing, Vol. 20, No. 6, June 2011. [4] Marina E. Plissiti and Christophoros Nikou, “ Overlapping Cell Nuclei Segmentation Using a Spatially Adaptive Active Physical Model, "IEEE Transactions on Image Processing, Vol. 21, No. 11, November 2012. [5] Xin Qi, Fuyong Xing, David J. Foran,andLinYang,“Robust Segmentation of Overlapping Cells in Histopathology Specimens Using Parallel Seed Detection and Repulsive Level Set, "IEEE Transactions on Biomedical Engineering, Vol. 59, No. 3, March 2012. [6] Chiwoo Park, Jianhua Z. Huang, Jim X. Ji, “ Inference, and Classification of Partially Overlapping Nanoparticles, "IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 35, No. 3, March 2013. [7] Sahar Zafari, Tuomas Eerola, Jouni Sampo, “ Segmentation of Overlapping Elliptical Objects in Silhouette Images, "IEEE Transactions on Image Processing, Vol. 24, No. 12, December 2015.