The research paper published by IJSER journal is about CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image 1

ISSN 2229-5518

CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image

Dr. Ali Hassan Al-Fayadh

Assistant Professor

Hind Rostom Mohamed

Assistant Professor

Raghad Saaheb Al-Shimsah

Master student

Mathematical Department

Computer Department

Mathematical Department

Mathematical and Computer

Mathematical and Computer

Mathematical and Computer

Sciences College,

Sciences College,

Sciences College

Kufa University, Iraq

aalfayadh@yahoo.com

Kufa University, Iraq

hindrustum.shaaban@uokufa.edu.iq

Kufa University, Iraq

Akram_00002002@yahoo.com



AbstractIn the present paper, Mean Shift Algorithm and active contour to detect objects for CT Angiography Image Segmentation is proposed.Based on the results we believe that this method of boundary detection together with the mean -shift can achieve fast and robust tracking of the CT Angiography Image Segmentation in noisy environment. The proposed scheme has been tested successfully on a large set of images. The performance of the proposed detector compares favorably both computationally and qualitatively, in comparison wit h Mean Shift and contour detector which are also based on surround influence .The last stage is stage contain Extraction of c onnected components CT Angiography image edge detection . The proposed scheme can serve as a low cost preprocessing step for high lev el

tasks such shape based recognition and image retrieval. The experimental results confirm the effectiveness of the proposed algorithm.

Key wardImage Segmentation,image detection, Image ,CT Angiography Image , mean -shift , contour Image.

1 INTRODUCTION

—————————— ——————————
A segmentation could be used for object recognition, occlu- sion boundary estimation within motion or stereo systems, im- age compression, image editing, or image database look-up[3].
The goal of image segmentation is to cluster pixels into sa- lient image regions, i.e., regions corresponding to individual surfaces, objects, or natural parts of objects.
In general these methods have not performed satisfacto- rily for image data due to their reliance on an a priori parame- tric structure of the data segment, and/or estimates of the num- ber of segments expected. Mean shift‘s appeal is derived from both its performance and its relative freedom from specifying an expected number of segments. As we will see, this freedom has come at the cost of having to specify the size (bandwidth) and shape of the influence kernel for each pixel in advance[1]. The application of the mean shift algorithm to color image seg- mentation has been proposed in 1997 by Comaniciu and Meer [2]. Since then it has become a widely used method for color image segmentation, as it provides significantly better segmen- tation results as other [3].
The mean-shift algorithm is an efficient
approach to tracking objects whose appearance is defined by
color. (not limited to only color, however. Could also use edge
orientations, texture, motion)[4].
The basic idea in active contour models or snakes is to evolve
a curve, subject to constraints from a given image , in order to
detect objects in that image. For instance, starting with a curve
around the object to be detected, the curve moves
toward its interior normal and has to stop on the boundary of the object[5].
Edge detection of an image reduces significantly the amount of data and filters out information that may be regarded as less relevant, preserving the important structural properties of an image. Therefore, edges detected from its original image contain major information, which only needs a small amount of memo- ry to store. The purpose of detecting sharp changes in image brightness is to capture important events and changes in prop- erties of the world [4].
Contour detection in real images is a fundamental problem in many computer vision tasks. Contours are distinguished from edges as follows. Edges are variations in intensity level in a gray level image whereas contours are salient coarse edges that belong to objects and region boundaries in the image[8].
the scheme is computationally expensive and produces a contour map which is quite sparser than an edge map though
not as sparse as the ground truth (contour map).
The paper is organized as follows; Section 2,deals with the
Mean Shift Algorithm segmentation. Section 3 deals with the
active contour to perform the connected component edge image
analysis, section 4 deals with the gives the overview of algo-
rithm with results and last section 5 ends the paper with conclu-
sion.

IJSER © 2012

http://www.ijser.org

The research paper published by IJSER journal is about CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image 2

ISSN 2229-5518

2 Mean Shift Algorithm

Segmentation is subdividing an image into its constituent re- gions or object. The level up to which the subdivision is carried out depends on the problem being solved. [5].
The mean shift algorithm is a nonparametric clustering tech- nique which does not require prior knowledge of the number of clusters, and does not constrain the shape of the clusters.
Given n data points xi, i = 1, ..., n on a d-dimensional space Rd,
the multivariate kernel density estimate obtained with kernel
K(x) and window radius h is:

(1)

For radially symmetric kernels, it suffices to define the profile of the kernel k(x) satisfying:

(2)

where ck,d is a normalization constant which assures K(x) inte- grates to 1. The modes of the density function are located at the
zeros of the gradient function f(x) = 0. The gradient of the
density estimator (1) is:
Represent color distribution with a histogram. Use mean-shift to find region that has most similar distribution of colors.

3 Active contour Image

Image editing tasks normally involve one or more extended contours, not single edge elements in isolation. For this reason, contour-based image editing depends upon an efficient method for specifying a group of edges to which an action is to be ap- plied. An efficient method for grouping edges into closed con- tours has recently been reported . The algorithm consists of three
main stages:
1. Line segment approximation.
2. Computation of posterior line grouping probabilities.
3. Shortest path computation of maximum-likelihood line seg-
ment cycles[7].

4 Experimental Results

In this section a detailed experimental comparison of the above stated algorithms has been presented. We have used two color image databases:
(1) CT Angiography Images database prepared in our condi-
tions ,images obtained from in thee Hospital.
(2) CT Angiography Images database obtained from internet.
Figure(1) shown sample data base for CT Angiography Images we used in this paper.

(3)
where g(s) = −k ‗ (s). The first term is proportional to the density estimate at x computed with


kernel G(x) = cg,dg( x 2) and the second term:

(4)

is the mean shift. The mean shift vector always points toward the direction of the maximum increase in the density. The mean shift procedure, obtained by successive.
• computation of the mean shift vector mh(xt),
• translation of the window xt+1 = xt +mh(xt)
is guaranteed to converge to a point where the gradient of den- sity function is zero. Using Mean-Shift on Color Models
Two approaches:
1) Create a color ―likelihood‖ image, with pixels.
2)weighted by similarity to the desired color (best for uncolored
objects)

Figure(1):

sample data base for CT angio graphy Images.

As we have seen, segmentation involves finding salient re- gions and their boundaries. A boundary in an image is a con- tour that represents the change from one object or surface to

IJSER © 2012

http://www.ijser.org

The research paper published by IJSER journal is about CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image 3

ISSN 2229-5518


another. This is distinct from image edges, which mark rapid changes in image brightness, but may or may not correspond to salient boundaries.
The scale of the mean-shift kernel roughly controls the size and shape of the extracted regions. There is a trade-off between maintaining the salient boundaries but suffering over- segmentation, versus missing some of the important boundaries and under-segmenting the image. Figure(2) shown contour for sample for CT Angiography Images. The segmentations above illustrate a typical compromise.
The resulting image that we obtained after Mean Shift Algo-
rithm segmentation would still contain some noise, which is made up of scattered image pixels and maybe some arbitrary
pixels of other objects that share similar tones to that of the im- age. It is also possible that some pixels are missing within re- gions of image because the segmentation was too strict, thus removing some pixels which are actually real skin. We end up with a much cleaner image after performing these operations. The subsequent step will perform various feature checks and gradient matching to finally confirm whether or not a particular region is image .

Figure(2): (A) Original CT Angiography Image , (B) Contour for Image

The mean shift algorithm seeks modes or local maxima of densi- ty in the feature space.
Figure(3) shown Appling mean shift algorithm for sample for
CT Angiography Images.

IJSER © 2012

http://www.ijser.org

The research paper published by IJSER journal is about CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image 4

ISSN 2229-5518


Where variable conn can have Value for Two-dimensional connectivity:
A) 4 if 4-connected neighborhood
B) 8 if 8-connected neighborhood
The 1-valued elements define neighborhood locations relative to
the central element of conn. Note that conn must be symmetric
about its central element.
The basic steps for the desired connectivity are:
1- Determine the connected components:
CC = bwconncomp(BW, conn);
2- Compute the area of each component:
S = regionprops(CC, 'Area');
3- Remove small objects: L = labelmatrix(CC);
BW2 = ismember(L, find([S.Area] >= P));
The next step we find connected components in binary
image .The basic steps in finding the connected components
are:
1- Search for the next unlabeled pixel, p.
2- Use a flood-fill algorithm to label all the pixels in the con-
nected component containing p.
3- Repeat steps 1 and 2 until all the pixels are labeled.

For instance, we could associate the pixels on either side of a contour with regions, identifying the ―figure‖ as the more con- vex side Figure (4) shown first and Second Connected Com- ponent then shown CT Angiography Image Detected Output.

Figure(3) : Appling mean shift algorithm for sample for CT Angiography

Images

5 Conclusion

We proposed a new goodness criterion for segmenting closed figures. We demonstrated our approach for Mean Shift Algo- rithm and contour Image based segmentation, but it should be straightforward to extend it to incorporate other features.
The demonstration task is that of finding a figure region edge
by contour distribution on the surrounding background. Our
approach is region based and applies most naturally to closed contours. We can extend it to open contours as long as the con- tour can be considered as organizing the image into distinct regions. We Connected Components for mean shift results by remove small objects and connected the connected components for skin image that have fewer than P pixels, producing another binary image .The default connectivity is 8 for two dimensions. We used the statement BW2 = bwareaopen(BW, P, conn) to spe- cifies the desired connectivity Figure 9 shown desired connec- tivity.

.

Figure 4 : (A) Original CT Angiography Image,(B) First Connected Com- ponent ,(C)Second Connected Component ,(D) CT Angiography Image Detected Output.

Finally, the object tracking process performs as memory for collecting skin-color objects obtained from previous frame to guide the next frame in order to remove image-color pixels that

IJSER © 2012

http://www.ijser.org

The research paper published by IJSER journal is about CT Angiography Image Segmentation by Mean Shift Algorithm and Contour with Connected Components Image 5

ISSN 2229-5518

immediately appear from frame to frame. The experimental results show the satisfying subjective test results.

6 References

6.1. Journal Article

[1] Prakash PANDEY, Uday Pratap SINGH and Sanjeev JAIN, Categorization and Searching of Color Images Using Mean Shift Algorithm, Leonardo Journal of Sciences, Issue 14, January-June (2009).

[2] C.NagaRaju, S.NagaMani, G.rakeshPrasad,S.Sunitha ,Morphological Edge Detection Algorithm Based on Multi-Structure Elements of Different Direc- tions, Volume 1 No. 1, International Journal of Information and Communica- tion Technology Research ,2010-11 IJICT Journal. All rights reserved , (2011).

6.2. Book

[3] Jue Wang, Bo Thiesson, Yingqing Xu, Michael Cohen, Image and Video Seg- mentation by Anisotropic Kernel Mean Shift, Proc. IEEE Int. Conf. on Com- puter Vision,,(2002).

[4] Farhad Dadgostar1, Abdolhossein Sarrafzadeh1, Scott P. Overmyer2,Face Tracking Using Mean-Shift Algorithm: A Fuzzy Approach for Boundary De- tection in IEEE Conference on Computer Vision and Pattern Recognition, (2005).

[5] Tony F. Chan, Member, IEEE, and Luminita A. Vese, Active Contours Without Edges, IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 10, NO. 2, FEBRUARY (2001)

[6] James H. Elder, Member, IEEE, and Richard M. Goldberg, Image Editing in the Contour Domain, IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 23, NO. 3, MARCH (2001).

[7] Gopal Datt Joshi and Jayanthi Sivaswamy ,A SIMPLE SCHEME FOR CON- TOUR DETECTION, In Computer Vision and Image Understanding.,(2002).

6.3. Conference Proceedings

[8] Quan Huynh-Thu Mitsuhiko Meguro Masahide Kaneko, Skin-Color-Based Image Segmentation and Its Application in Face Detection, IAPR Workshop on Machine Vision Applications, (2002),Dec. 11 – 13. Nara- ken New Public

Hall, Nara, Japan.

IJSER © 2012

http://www.ijser.org