You are on page 1of 17

BACKTRACKING SEARCH

ALGORITHM FOR COLOR IMAGE


MULTILEVEL THRESHOLDING
1 . S . P a r e 1 ; 2 . A . K . B h a n d a r i 1 , 3 ; 3 . A . K u m a r 1 , 2 ; 4 . V.
Bajaj1

PRESENTED BY:--
Mst. Sonali Akter Swarna-2040
Rasedul Mannan-2060
Mahmodol Hasan Monna-2076
Problem Statement :

Find a solution that can


disjunct the object &
background to effectively
deal with the natural or
satellite image.

Add a Footer 2
Designing a MFE-BSA algorithm

Modified Fuzzy Entropy


Backtracking Search
Algorithm.

Add a Footer 3
OBJECTIVES:
 Removing fuzziness & ambiguity
from natural & satellite images.

 Providing thresholding levels in all


regions having almost equal
entropies.

 Using BSA improve the


performance of MFE.

 Presenting robustness & stability of


MFE-BSA in overall segmentation
results.

Add a Footer 4
LETS KNOW ABOUT THRESHOLDING

Thresholding is a type of image segmentation. The simplest


thresholding methods replace each pixel in an image with a black
pixel if the image intensity or a white pixel if the image intensity
is greater than that significant value.

Multilevel thresholding is a process that segments a gray level


image into several distinct regions.

Add a Footer 5
o Multilevel thresholding

o Metaheuristic algorithms

o Modified fuzzy entropy

o Backtracking search algorithm

o Image detection

Add a Footer 6
 Backtracking Algorithm
has 5 steps. Following
they are:

 Initialization
 Selection - 1
 Mutation
 Crossover
 Selection - 2

 Membership function:

 Membership function is a
user-defined function that
define the float values
between 0 and 1 from fuzzy
logic.
7
TITLE
1.Multilevel thresholding
2. Statistical Equation
algorithm on BSA
 Step 1 Extract the red, green and blue pixel matrixes from
an input image.
o Entropy
 Step 2 For each of the individual image matrixes, obtain the
(Measure of randomness)
histogram distributions, and then set the number of discrete
levels for the image to be segmented.
o MSE
 Step 3 Set the required control parameters (N, D, mixrate,
low1:D, up1:D, maximum iterations) for BSA. (Assesses the average squared
 Step 4 Input image histogram and number of thresholds.
difference between the observed
 Step 5 Initialize the population for basic components of
color image (red, green and blue) using randomly chosen and predicted values.)
gray-level pixel intensity values between 0 and 255.

 Step 6 For each of the color components, form the required


number of classes from the corresponding position of every
particle in population P. Form each of the class by using pre-
defined number of threshold values as pixel intensity from
image.

 Step 7 Obtain best fuzzy parameters combination using BSA


algorithm that can minimize the fitness function.

 Step 8 Determine the threshold values and the


corresponding segmented image.

Add a Footer 8
Trapezoidal membership function equation:

Fuzzy Entropy function equation:

Minimum Fuzzy parameter entropy of


Entropy of adjacent Threshold division: individual region:

9
PSNR(Peak Signal to Noise Ratio):

MSE(Mean Square Error):

Add a Footer 10
CPU(Computing Processing Time):

SSIM(Structural Similarity Indexing):

Add a Footer 11
FITNESS VALUE
Energy-Tsalli’s-CS < Tsalli’s-CS < MFE-BFO < MFE-BSA
And on increasing CPU time like Tsalli’s-CS < MFE-BSA < MFE-BFO <
energy-Tsalli’s-CS.

If we compare 3/5/8/12 level thresholding we see that segmentation is


much smoother and more uniform at higher threshold levels using MFE-
BSA.

12
DISCUSSION

• Best PSNR and Minimum MSE


• SSIM is closer to unity
• Average Computational time
• Finally we get an approach for multilevel thresholding
using MFE function & BFA algorithm which meets
with a suitable values for every parameter
(PSNR,MFE,SSIM,FSIM). After all these, we get the
most accurate interpretable image.

Add a Footer 13
Any Drawback?

Add a Footer 14
RELATED WORK

1.Deep Learning 2.Computer


vision (Open CV)

Add a Footer 15
REFERENCE
1. Chang, C.C., Wang, L.L., 1977. A fast multilevel thresholding method based on low pass and
high pass filtering, Pattern Recognition Letters 18
2. Liao. P.S., Chen, T.S., Chung, P.C., 2001. A fast Algorithm for Multilevel Thresholding. Journal
of Information Science and Engineering
3. Zhang, Y. (2011). "Optimal multi-level Thresholding based on Maximum Tsallis Entropy via an
Artificial Bee Colony Approach". Entropy. 13 (4): 841–859. Bibcode:2011Entrp..13..841Z.
doi:10.3390/e13040841.
4. https://www.youtube.com/watch?v=L-sFn4EnObA.
5. Pun, T., 1980. A new method gray-level picture thresholding using the entropyof the histogram.
Signal Processing
6. Whatmough, R.J., 1991. Automatic Threshold Selection from a histogramusing the exponential
hull. Graph, Models Image Process.
7. K. Hammouche, M. Diaf and P. Siarry, "A multilevel automatic thresholding method based on a
genetic algorithm for a fast image segmentation", Comput. Vis. Image Understand., vol. 109
8. A. K. Bhandari, A. Kumar, S. Chaudhary and G. K. Singh, "A novel color image multilevel
thresholding based segmentation using nature inspired optimization algorithms", Expert Syst.
Appl., vol. 63
9. Wu, S., Amin, A., 2003. Automatic Thresholding of Gray-level Using Multi-stage Approach,
ICDAR 38
10. .https://www.youtube.com/watch?v=DcWrbsPJEd8

Add a Footer 16
THANK YOU

Add a Footer 17

You might also like