1. DANGER DETECTOR FINAL PRESENTATION
Aarthi Balachander
Krithika Chandrasekar
Govind Manian
Charvaka Mattaparthy
Aziza Satkhozhina

2. Objectives
The algorithm takes as its input a CT image of screened baggage, each image comprised of multiple slices. Each slice represents a 2D view of the outline of each object contrasted against a background. Our immediate objective is:
To find a suitable method to separate 2D objects and background. Several thresholding techniques can be used (Otsu method helps to separate data into two or more classes of pixels.)
To perform connected component analysis on the slices to obtain the 2D connected object.
Perform clustering (or another suitable segmentation algorithm) if there is a distinct change in density across the volume of the object and confirm the exact region around which an explosive might be concealed.

3. Flowchart of algorithm

4. Otsu’s Method

5. Otsu Thresholding
Use of graythresh command in Matlab to find a threshold using Otsu’s Method

6. CONNECTED COMPONENT ANALYSIS
Once region boundaries have been detected, it is often useful to extract regions which are not separated by a boundary. Any set of pixels which is not separated by a boundary is call connected. Each maximal region of connected pixels is called a connected component. The set of connected components partition an image into segments.
Let s be a neighborhood system.
– 4-point neighborhood system
– 8-point neighborhood system
• Let c(s) be the set of neighbors that are connected to the
point s.
For all s and r, the set c(s) must have the properties that
– c(s) ε s
– r ε c(s) , s ε c(r)
A region R S is said to be connected under c(s) if for all s, r 2 R there exists a sequence of M pixels, s1, · · · , sM such that
s1 ε c(s), s2 ε c(s1), · · · , sM ε c(sM−1), r ε c(sM)
i.e. there is a connected path from s to r. *

7. EXAMPLE 1 s = (i, j); ClassLabel (0, 0); 1 The segmentation Y i
The image X j i

8. EXAMPLE 2 s = (i, j); ClassLabel (0, 1); 2 The image X j 0 1 2 3 4
The segmentation Y j i

9. EXAMPLE 3 s = (i, j); ClassLabel (2, 0); 3 The image X j 0 1 2 3 4
The segmentation Y j i

10. EXAMPLE 4 s = (i, j); ClassLabel (4, 4); 4 The image X j 0 1 2 3 4
The segmentation Y j i

11. Connected Component Analysis

12. Region labeling algorithm
Relative Performance in different development environments
*Performance results for Intel Pentium 2.4 MHz Processor

13. Feature Vector analysis
Mean and Variance are the numerical features that are used to represent our object
str1 = 'mean';
str2 = 'var';
eval([featvector '= struct( str1,mean(mean((A))), str2, var(var((A))));']);
eval([featvectorOriginal '= struct( str1,mean(mean((C))), str2, var(var((C))));']);
x = mean(mean((A)));
y = var(var((A)));
xOriginal = mean(mean((C)));
yOriginal = var(var((C) ) );
meanGray(i) = x;
varGray(i) = y;
meanOrig(i) = xOriginal;
varOrig(i) = yOriginal;

14. Hierarchical Clustering

15. Hierarchical Clustering

16. Hierarchical Clustering

17. Hierarchical Clustering

18. Hierarchical Clustering

19. Hierarchical Clustering

20. Hierarchical Clustering

21. Hierarchical Clustering

22. Hierarchical Clustering

23. Hierarchial clustering
In hierarchical clustering the data are not partitioned into a particular cluster in a single step. Instead, a series of partitions takes place, which may run from a single cluster containing all objects to n clusters each containing a single object.  Hierarchical Clustering is subdivided into agglomerative methods, which proceed by series of fusions of the n objects into groups, and divisive methods, which separate n objects successively into finer groupings.
for cnt = 1:20
if (cnt>1)
for n = 1:m
h=min(rowmin,colmin) ;
hnot = max(rowmin,colmin) ;
dist(h,n) = min(dist(rowmin,n), dist(colmin,n));
dist(n, h) = min(dist(n,rowmin), dist(n,colmin));
end
dist(hnot,:) = [];
dist(:, hnot) = [];
m = m-1;
minv = 10000;

24. Classification
if (d==1) if ((abs (xOriginal - Feat1 (1))<1)||(abs (yOriginal - Feat1 (2))<1)) fprintf ('Location is found! Slice = % d Object # = % d \n',k,i) disp (row') disp (col') end if ( c == 1) Feat1 = [xOriginal yOriginal]; d=d+1; c=c+1;

25. Future work
Perform connected component analysis and follow it up by 3D region growing across multiple slices to obtain the 3D connected object.
Perform segmentation in 3D for more accurate analysis.
Detect False alarms using MAP estimation and accurately find co-ordinates of the explosive, given the 2D CT slices of scanned baggage