IMPROVING IMAGE RESOLUTION THROUGH THE CRA ALGORITHM INVOLVED RECYCLING PROCESS TO IMAGE RECONSTRUCTION

David C. Wyld (Eds) : ICCSEA, SPPR, CSIA, WimoA - 2013
pp. 263–272, 2013. © CS & IT-CSCP 2013 DOI : 10.5121/csit.2013.3527
IMPROVING IMAGE RESOLUTION
THROUGH THE CRA ALGORITHM
INVOLVED RECYCLING PROCESS TO
IMAGE RECONSTRUCTION
S. N. Tazi1
, Deepika Verma2
, Vinesh Jain3
, A. K. Dubey4
, Rakesh Rathi5
1, 2
M.Tech (Student) CSE & IT
3,4,5
Assistant Professor (CSE Dept)
1,4
Member IEEE
Govt. Engineering College Ajmer-305001, Rajasthan (INDIA)
1
satya.tazi@gmail.com, 2
deep.verma834@gmail.com,
3
vineshjain1280@gmail.com, 4
aniludenish@ieee.org
5
rathirakesh4@rediffmail.com
ABSTRACT
Image processing concepts are widely used in medical fields. Digital images are prone to a
variety of types of noise. Noise is the result of errors in the image acquisition process for
reconstruction that result in pixel values that reflect the true intensities of the real scenes. A lot
of researchers are working on the field analysis and processing of multi-dimensional images.
Work previously hasn’t sufficient to stop them, so they continue performance work is due by the
researcher. In this paper we contribute a novel research work for analysis and performance
improvement about to image resolution. We proposed Concede Reconstruction Algorithm (CRA)
Involved Recycling Process to reduce the remained problem in improvement part of an image
processing. The CRA algorithms have better response from researcher to use them.
KEYWORDS
image, noise, acquisition, reconstruction, CRA, visualization.
1. INTRODUCTION
Visual perspective of object is the current theme for computers device to make them more
effective for vision. There is the potential to sense body position, head orientation, direction of
gaze, pointing commands, and gestures. Interactive vision is continuous growth to the computer
application. A game player didn’t think to press the button every time (or they didn’t want to
directly connect the computer physical want to directly connect the computer physically), they
required automated vision system through that sensor automatically sense out the activities
(action) movement done by game player is recognised. A CAD designer may be tiered due to the
labour work for physically connect to computer for manipulate object through their finger and
hands. Similarly different machine application user is also suffering too many types of action
listener. The Concept of computer vision and human computer interaction make the machine

264 Computer Science & Information Technology (CS & IT)
interaction safety, joyful and interesting. They reduce lot of risk and few challenges occurred to
achieve their motive. Their application response time is touch the user vision thought. The
algorithm for computer vision is designed to create the base line for bulk people reliability of
their vision and also reduce the cost factor of existing system.
Now a day’s researcher and academic both are emerging to highlight the previous work of image
processing and search novel inventory for recon sting the user action. To catch the exact action of
user is more difficult and their reorganisation, reconstruction is so complex. There for lot of
researcher are involving to tackle to this goal but can’t achieve so continues research is going
now. The vision input function as part of larger system that include the interactive graphics
display. They mainly focus to control different application such as computer game, television
program graphically they control and show to the user that objects exist in the real world. Here
we focus on various fundamental of visual measurement. In incremental order of complexity
large object tracking reorganisation of shape edges and motion analysis of object to track them.
We are proposed to different approaches for tracking the object and maximise the reorganisation
of object [1].
1.1 Digital images Vision
The representation of image object in two dimensional grid is in the form of row * column (size
m*n, where m for row & n for column). m* n shown the pixel of images which have a value to
encode the colour and intensity of image. E.g. the RGB colour image pixel representation in
triplet [R; G; B]. Respectively Red, Green & Blue colour repenting the gray level image. In pixel
form contain single value corresponding to their intensity [2].
• Input Image: The initial phase of image processing is image acquisition. The sensors of
kinetic camera sense-out the image object in different projection and collect all intensity
of image data.
• Pre-processing: The systematically arrangement of multiple captured image data is
known as pre-processing stage. This phase also cover the manipulation (e.g.
enhancement) and reduction of noise.
• Segmentation: In this phase we categorise the whole image in many parts such as edge
and region of images. So that expected changes might be done accordingly needed.
• Reconstruction of data: the edge and region of image data is converted from series of m-
dimensional to n-dimensional. Both dimension M and N are matched to their region with
edge.
• Matching: the registration of all n-dimensional data are putted in an order so that, the
combination made useful information. This method is applying on data from
distinguishing sources or single source applied to different times.
• Recognition: The recognition of object does not achieved in a signal phase of
identification of element that may depend on the final purpose of the analysis.
The pipelining of computer vision & image analysis pipeline is as follows:-

Computer Science & Information Technology (CS & IT) 265
Fig.1: Generic image analysis and computer vision pipeline [3]
2. PROJECTIVE CAPTURING TECHNIQUE
2.1 Calibration
A most common projective capturing technique is camera calibration. It involves the estimation
of both side camera parameter such that extrinsic and intrinsic. The calibration of multiple views
set up two cameras for computing different parameter of object (basic calibration algorithm).
Camera parameter estimation is a practice varies approach to calibrating the employed object.
The correction of lens distortion required calibration rig (known as geometry), to estimate the
camera based parameter object. i.e., check board pattern. The position and it’s orientation of
internal parameter camera is calculated through the different perspective view of check board
using calibration algorithm. The strong calibration of camera object is a process to compute
different view calibration of camera parameter which must be addresses as follows:-
1. Capture N (at least 3) images with the checkerboard pattern and estimate point-
correspondences.
2. Estimate and correct the radial lens distortion.
3. For each captured image, calculate the N homographic transforms.
4. Using the N homographic transforms, calculate the intrinsic and extrinsic parameters.
5. Refine the calculated camera parameters as explained in Section D above.
Input image
Pre-Processing
Segmentation, Features Extraction
Reconstruction
Matching
Recognition

The correction method of non-linear radial lens distortion for calibration process given more
accrued result to obtain homographic Hi in compared to standard technique. This we consider
(hm
)T
= hmT
, where first one represent the multiplication of inner product and another one is a
equation to lead a scalar.
2.2 Epipolar Geomatry
To estimate the depth information simulate according to human eyes. We required stereo vision
simulation technique. As we know that real world object is projected into 2D coordinate. Through
camera real world projection technique describe as two point X and Y in the given figure the
projection of point X is pointed out the object surface. Space point X of object is projected in 2D
coordinate by using of two different cameras C1and C2 which pointed out x and x′ our projection
point to respective camera. The line that connects C1 and C2 points is baseline. The point’s e and
e′ on the baseline are epipolar poles. Any point in real world space compose different epipolar
plane with two projection plane. But the baseline and epipolar poles were never changed. The
lines crossed by the epipolar and projection planes are epipolar lines l1 and l2.
Fig. 2 Image Processing 1) Left image 2) Right image [5]
Any point on the epipolar plane will be projected on the epipolar lines l1 and l2. According to this
rule we can reduce the researching area to find out the projection point the two images.
3. FILTERING TECHNIQUE
Noise: anything that make changes in the vision of original images also known as image noise.
To remove noises from those images is the part of filtering technique. By the use of this classical
and novel filtering technique we achieved the nearest copy of original image. Few of common
filtering technique are preferred for image noise as follows:-
(i) Low pass filtering
The simplest and easiest method of filtering to decrease the image noise is known as low pass
filtering .This method is applying to calculate the average image noise and minimal filter to
removal them. For the work of minimal filtering it’s called Gaussian or average filtering.

(ii) High pass filtering
In the case of minimal filtering the sharpness of image can’t recover its original form (At the
blurred and complex image noise). To reduce such a problem we use the maximum filtering to
reconstruct the originality of image noise. This is known as high pass filtering method.
The above both filtering method are the part of linear operation taken at Image noise. In the case
of Non-linear operations other filtering method such as median or out-of-range filters are applied.
A special class of algorithm for low pass filtering is applied to enhance the object desirable.
Consider many degraded and translated and rotation of object is available for particular
application. The idea behind these motive are called correlation averaging which align the
elements in two given standard as correlation technique to estimate the internal relation between
objects and another one is the calculate the average of that object correlation. The special class of
resolution of images are based on signal-to-noise ratio. The resolution indirectly grows up on the
basis of square root of the number of averaged individual element. On the basis of distortion of
images relevant correction are required for example Geometric distortion call to relevant
Geometric correction that is the necessity of image registration.
4. RECONSTRUCTION
4.1 Basic Concept of Reconstruction
Let us suppose that we want to share our experience and wonderful momentum to our friends and
relatives, and then we capture the images in different angles and shared to them. The principle
behind capturing different angle images is occurred in our mind cause of distortion of single
views image. These thinking phenomena of different angle images captured by camera generate
the basic concept of reconstruction.
A systematic projection of acquired images for reconstruction as users view shown in figure
below:-
Fig.3 Projection data acquired from different views is used to reconstruct the image. [4]
4.2 Definition of Reconstruction
Geometric reconstruction is defined as construction of damaged images and multiple views of
images fusion from rebuild to redesign. An acquired view of images categories the reconstruction
in different terms.
These terms are follows:

(1)Substitutions
(2) Typing with constraint sets (type equation)
(3)Unification solving constraint sets.
(4) Principle types
(5) Let Polymorphism
4.3 Basic Algorithms for Reconstruction
1) Filtered back-projection algorithm: “A standard mathematical procedure to distribute the equal
amount of activities in every section of projection for every angle, to reconstruct the original
image for spike back technique is known as back projection algorithm“. Through the help of back
projection algorithm we get blurred version of original image.
Proposed Approach:-
Let us assume that
б B (I) =C=̃A. (1)
A= original image.
B= Back projection Method
C= Reconstructed image.
I = Noise image.
б is standard constant depending on angle/degree rotation.
For a maximum reconstruction
б is nearest to 180˚.
2) Analytical algorithms: In this algorithm, the Fourier transform and its inverse are taken for the
iteration. These algorithms use the basic concept of back projection algorithm.
3) Three-dimensional image reconstruction: The formation of three-dimensional (3D) image is
constructed through the help of 2D images slices. The access plan measurement is composed by
both back projection filtered and filtered back-projection algorithm.
4) Iterative reconstruction: The iterative reconstruction algorithm is most popular in nuclear
medicine because of following reasons:-
(a) For the low counts the algorithm is simple to model and manage projection noise.
(b) Simply modelled the imaging physics (i.e., non-uniform attenuation, geometry, scatter etc.)

Fig.4. Flow chart of iterative image reconstruction scheme. [4]
5) Eight-point linear algorithm: For creating essential matrix in estimation of computer vision,
eight point algorithms are used. Appropriate given matrix shown the stereo vision camera pair to
the corresponding point of image.
5. CONCEDE RECONSTRUCTION ALGORITHM
5.1 Proposed CRA
Step1. Capture bundle of images from multiple cameras
Step2. Pre-Processing
Step3. Apply Noise Removing Technique
3.1 if noise ==Gaussian noise(salt-and-pepper or speckle)
Removing Noise By Linear Filtering
Else if noise==blur
3.2 Removing Noise by Adaptive Filtering
Else
3.3 Removing Noise by Median Filtering
Step4. Apply any segmentation techniques
4.1 Clustering methods
4.2Compression-based methods
4.3 Histogram-based methods
4.4 Edge detection
4.5 Region-growing methods
4.6 Split-and-merge methods
4.7 Partial differential equation-based methods
4.8 parametric methods
4.9 Level set methods
4.10 Fast Marching methods
4.11Graph partitioning methods
4.12Watershed transformation and
4.13 Multi-scale segmentation

Step5. Partition of images at aspect levels
{
Complete Image/4=A,
A/4=B,
B/4=C,
…
…
Nth/4= æ (aspect level)
}
Step6. Render æ images
Step7. Reconstruct render images
Nth
Rc = ∑ æ
0
Step8. Compare reconstructed image from original image
Rc Ro
Step9. if noise is occurred
Rc ≠ Ro
Then Backtracking step 3
Else step 10
Step10.Exit
In this algorithm we are capturing a bundle of image from multiple cameras and applying per-
processing technique to give the systematic arrangement and applying noise removal techniques
for enhancement of image. For applying CRA algorithm we are classified region and edges of an
image. After found of edges and region we are dividing whole image by four this process do
continuously until user accepted level is achieved. The rendering process is applicable to those
achieved object. The render object of image is walking through the reconstruction techniques.
After reconstruction of an image compared have to original image. Recycling this process until
comparison made equivalent.

5.2 Flow Chart of CRA
Fig. 5 Proposed Model of CRA
6. CONCLUSION
The graphics, animation and visualization are most important terms used in daily life. HCI
involve the better human vision through improvement of visualization of picture /images. We
study the different terms in image processing from historical perspective and find out there
projective capturing technique .i.e, calibration and epipolar geometry are commonly used. Two
filtering techniques (low pass and high pass) are used to remove the noise from blurred image.
For the improvement of image visualization many reconstruction algorithm are used. all of these
technique based on back projection algorithm. The available algorithm does not provide image
visualization at aspect level. To remove such a problem we proposed a novel algorithm Title as
“Concede Reconstruction Algorithm (CRA)”. The CRA algorithm improved the visualization of
image object with aspect ratio.

ACKNOWLEDGMENTS
The author wishes to thank faculty staffs for helpful comments and for help in preparing the
manuscript. They thank his class-met for encouragement and support at GECA. Document is
drawings &prepared by Sh. Anil K Dubey , S.N. Tazi& Deepika Verma.
REFERENCES
[1] M. Roth, K. Tanaka, C. Weissman, W. Yerazunis,. Computer Vision for Interactive Computer
Graphics, TR99-02 January 1999, Mitsubishi Electric Research Laboratories.
[2] Daniel Vaqueroa, Matthew Turka, Kari Pullib, Marius Ticob, Natasha Gelfandb, A survey of image
retargeting techniques, Nokia Research Center 955 Page Mill Road Palo Alto, CA, 94304-1003, USA,
SPIE 7798, 779814 2010.
[3] Thierry Pun, Guido Gerig and Osman Ratib, Image Analysis and Computer Vision in Medicine, into
Computer vision in medicine, February 8, 1993.
[4] G.L. Zeng, Image reconstruction - a tutorial, Computerized Medical Imaging and Graphics 25 2001.
[5] Deepak Gaur, Yaduvir Singh, Raj Kumar Sagar, 3D Reconstruction of Uniform Curved Object Model
From 2D Images, International Journal of advancement in electronics and computer engineering
(IJAECE) Volume 1, Issue1, April 2012, pp.1-3, ISSN 2278 – 1412.

IMPROVING IMAGE RESOLUTION THROUGH THE CRA ALGORITHM INVOLVED RECYCLING PROCESS TO IMAGE RECONSTRUCTION

More Related Content

What's hot (18)

Similar to IMPROVING IMAGE RESOLUTION THROUGH THE CRA ALGORITHM INVOLVED RECYCLING PROCESS TO IMAGE RECONSTRUCTION (20)

More from cscpconf (20)

Recently uploaded (20)

IMPROVING IMAGE RESOLUTION THROUGH THE CRA ALGORITHM INVOLVED RECYCLING PROCESS TO IMAGE RECONSTRUCTION