1. The document summarizes the first Doctoral Committee (DC) meeting of M. Theodore Kingslin, an Assistant Professor in the Department of ECE at RMK College of Engineering and Technology.
2. The proposed research topic is "Medical X-Ray Image Enhancement Based on Variations of Perona-Malik Model with Different Edge Stopping Functions". Perona-Malik model is a nonlinear anisotropic diffusion technique to remove noise while preserving edges. Different edge stopping functions will be evaluated.
3. Performance will be measured using PSNR and MAE metrics calculated on denoised images. Preliminary results show the edge stopping function c3 achieves highest PSNR, best preserving sharp edges
Interfacing Analog to Digital Data Converters ee3404.pdf
THEODORE KINGSLIN_PRESENTATION_JULY_2023 (1).ppt
1. FIRST DC MEETING
1
Name of the Scholar : M.THEODORE KINGSLIN
Register No : 23149997282
Month & Year of admission: July 2023
Designation : Assistant Professor
Department : ECE
Affiliation : 4112807 - R.M.K. College of Engineering and
Technology
Date : 22.08.2023
2. 2
Name of the Supervisor : Dr. N.GANGATHARAN
Supervisor Reference Number : 1940081
Designation : Professor & Head
Department : ECE
Affiliation : 4112807 - R.M.K. College of Engineering and
Technology
FIRST DC MEETING
3. 3
Name : Dr. K. HELEN PRABHA
Designation : Professor & Head
Department : Electronics and Communication Engineering
Organization/Institute : R.M.D Engineering College
Name : Dr. A. BRINTHA THERESE
Designation : Professor
Department : School of Electronics Engineering
University/Institute : Vellore Institute of Technology Chennai
FIRST DC MEETING
4. FIRST DC MEETING
4
MEDICAL X-RAY IMAGE ENHANCEMENT
BASED ON VARIATIONS OF PERONA
MALIK MODEL WITH DIFFERENT EDGE
STOPPING FUNCTIONS
5. 5
Sl.
No.
Degree Specialisation University/ Institute
Year of
Passing
Marks
(%/CGPA)
1 B.E ECE
Government College of
Engineering / Anna
University
2009 63.56
2 M.E
Applied
Electronics
Rajas Engineering College
/ Anna University
2012 79
FIRST DC MEETING
6. 6
Sl.
No.
Designation Organisation From To
Years of
Experience
1 Lecturer The Rajas Engineering College 07-06-2012 28-06-2013 1 year
2
Assistant
Professor
Rajas Engineering College 29-06-2013 06-07-2022 9 years
3
Assistant
Professor
R.M.K College of Engineering &
Technology
15-07-2022 Till Date 1 Year & 1 Month
Total Years of Experience
11 Years &
01 Months
FIRST DC MEETING
7. 7
Sl.
No.
Title of the Paper Authors
Name of
the Journal
Year,Volume,
Issue
Impact
Factor as
per Clarivites
1
Augmented
Reality.based Smart
Home Automation
using Dynamic vision
Sensor’.
Chairma Lakshmi.K.R,
Praveena B,
Golden Stepha,
Theodore Kingslin M,
Gurumurthy J,
Vijaya Anandh R
IEEE Explore
2023, pp. 884-887,
doi:10.1109/ICIDC
A56705.2023.1009
9860.
--
2
AWSN-Based
Correlation assessment
for a multi target
recognition protection
system ‘s sleep
scheduling
machanism.
David Neels Ponkumar.D,
Vimala Josphine.C,
Narendra Kumar.A,
Theod0re Kingslin.M
European
Chemical
Bulletin
Volume-12,Issue-
8(2023)
--
FIRST DC MEETING
8. Prona-Malik model which is non-linear anisotropic diffusion model proposed by
Perona and Malik. It is a partial differential equation based image processing
technique.
Perona-Malik model can successfully remove noise while preserving edges and
small structures as long as the diffusion coefficient or edge stopping function
F( ∇, ) and gradient threshold parameter (K) are estimated correctly.
Since there are several edge stopping functions, their ability to filter the noise
and ability to preserve the edges when used in Perona-Malik model need to be
studied and evaluated to make an appropriate choice.
8
FIRST DC MEETING
9. To improve the visual quality of Medical X-Ray
Image, Based on Variations of Perona-Malik
Model with Different Edge Stopping Functions.
In Perona-Malik Model, PSNR and MAE are
calculated to measure the performance of
denoising.
9
FIRST DC MEETING
10. 10
Among the the edge stopping functions, PSNR values are high for model using
edge function c3
The sharp edges and fine details are well preserved hence its PSNR is high.
Sharp edges and fine details of my X-ray image will be preserved well by using
other edge stopping functions.
11. 11
Year Authors Title Methodology
2022 Rex Paolo C. Gamara
Pocholo James M. Loresco
Argel A. Bandala
Medical Chest X-ray Image
Enhancement Based on
CLAHE and Wiener Filter
for Deep Learning Data
Preprocessing
Upgraded X-ray
image enhancement
hybrid algorithm that
utilizes and consists
of the Contrast
Limited Adaptive
Histogram
Equalization
(CLAHE) method
combined with the
Wiener filter
FIRST DC MEETING
12. 12
Year Authors Title Methodology
2022 Megha Sharma
Dharmendra Kumar
Contrast Enhancement and
Noise Reduction of chest X-
ray images
The contrast
enhancement and
noise reduction, using
Histogram
equalization, CLAHE
(Contrast Limited
Adaptive Histogram
Equalization), median
filter and DCT filter
for chest X-ray
images.
FIRST DC MEETING
13. 13
Year Authors Title Methodology
2022 Lucky Indra Kesuma
Ermatita Ermatita
Erwin Erwin
Purwita Sari
Rudhy Ho Purabaya
Improved Chest X-Ray Image
Quality Using Median and
Gaussian Filter Methods
The contour improvement
technique to the image using
Morphology Opening, and
followed by noise reduction
using Median Filter and
Gaussian filter. The results of
the two methods of noise
reduction are compared with
the results of image quality in
order to find out the best
method that can be applied.
FIRST DC MEETING
14. 14
FIRST DC MEETING
For image denoising isotropic diffusion can be used.
Isotropic diffusion is modeled as a two dimensional partial differential equation
known as Heat Equation.
Denoising an image the heat equation can be solved at different instance of time t.
Heat equation is described as given in equation below:
Perona-Malik model is described as given in equation below:
Perona-Malik Model:
15. Three edge stopping functions are suggested:
15
FIRST DC MEETING
Other edge stopping functions are given below:
Tukey’s biweight function
Zhichang Guo edge stopping function
Weickert edge stopping function.
Edge stopping functions:
16. Calculating Gradient:
The image gradient is not a reliable measure since it is
influenced by noise:
To overcome this problem, replace the term
Where Gσ is a Gaussian filter of scale σ
16
FIRST DC MEETING
17. Denoising Performance Metrices:
The following metrics we calculated to measure the performance of denoising with
Perona-Malik model. The Peak signal-to-noise ratio (PSNR) and Mean absolute
error (MAE) are defined as:
where I and I0 are original image and reconstructed image and M:N represents the size
of the image respectively, in horizontal and in vertical direction
17
FIRST DC MEETING
PSNRgrad is used to measure how well derivatives of restored image match those of
the original image:
18. The implementation of Perona-Malik model with different
edge stopping functions in Medical X-ray enhancement can
successfully removes noise, well preserve the sharp edges
and fine details in the Medical X-Ray Image.
18
FIRST DC MEETING
19. 1. J.Weickert, Anisotropic diffusion in image processing, ser. ECMI Series. Stuttgart,
Germany:Teubner-Verlag. 1998.
2. Yan-Fei Zhao, Qing-Wei Gao, De-Xiang Zhang and Yi-Xiang Lu, Medical X-Ray Image
Enhancement Based on Kramer’s PDE Model, Journal of electronic science and technology of
china, vol.5, No.2, June 2007.
3. Black M.J., G.Sapiro, D.Marimont, D.Heeger, Robust anisotropic diffusion, IEEE Transactions
on Image Processing, March 1998, Vol 7, No 3.
4. LinaSeptiana, Kang-Ping Lin, X-ray Image Enhancement using a Modified Anisotropic
Diffusion, IEEE International Symposium on Bioelectronics and Bioinformatics, 2014.
5. Kui Liu, Jieqing Tan, Benyue Su, Adaptive anisotropic diffusion for image denoising based on
structure tensor, International Conference on Digital Home, 2014..
19
FIRST DC MEETING
20. 20
FIRST DC MEETING
Sl.
No.
Course
Code
Title of the Course Work No. Of Credits
Core course/ Elective/
Special Elective
1. 20CS121
Research Methodologies and Intellectual
Property Rights (SEM-1)
3 Humanities and Science
2. 20CS125 Machine Learning Techniques (SEM-1) 3 Core Course
3. 20CS951 Image Processing and Computer Vision (SEM-2) 3 Elective Course
4. 20CS222 Internet of Things (SEM-2) 3 Core Course
5. 20CS956 Cyber Security and Ethical Hacking (SEM-2) 3 Elective Course
6. 20CS973 Deep Learning (SEM-3) 3 Elective Course