18.11.2014 Views

Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab

Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab

Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.3 Structure-based classification<br />

4.3.2.2 Renumber<strong>in</strong>g the nodes<br />

As already mentioned above renumber<strong>in</strong>g is necessary s<strong>in</strong>ce we want a unique number for<br />

each node <strong>in</strong> the tree depend<strong>in</strong>g only on the path to the node. Normally the nodes <strong>in</strong> a tree<br />

would be numbered arbitrarily. By renumber<strong>in</strong>g the vertices of the quadtree we get a list of<br />

vertices which is def<strong>in</strong>ed very similar to equation (4.27):<br />

V = {v u1 , . . . , v un } with |V | = n (4.29)<br />

where the numbers u 1 , . . . , u n are the unique numbers of the respective nodes.<br />

These numbers can be mapped through a function U<br />

U : V −→ N (4.30)<br />

where U must have the properties<br />

and<br />

U(v i ) = U(v j ) ⇔ i = j (4.31)<br />

U(v i ) ≠ U(v j ) ⇔ i ≠ j (4.32)<br />

4.3.2.3 Unique number generation<br />

A node <strong>in</strong> a quadtree can also be described by the path <strong>in</strong> the tree lead<strong>in</strong>g to the specific<br />

node. As we saw <strong>in</strong> section 4.3.1, if a node <strong>in</strong> a quadtree has child nodes there are always<br />

exactly four children. This means that for each node <strong>in</strong> the tree hav<strong>in</strong>g child nodes we have<br />

the choice between four possible child nodes to traverse down the tree.<br />

By def<strong>in</strong><strong>in</strong>g the simple function s<br />

⎧<br />

1, if v l is the lower left child node of the parent node<br />

⎪⎨<br />

2, if v<br />

s(v l ) =<br />

l is the lower right child node of the parent node<br />

3, if v l is the upper left child node of the parent node<br />

⎪⎩<br />

4, if v l is the upper right child node of the parent node<br />

, (4.33)<br />

we can express the path to a node v i as sequence P vi which is def<strong>in</strong>ed as<br />

P vi = (p 1 , . . . , p M ) (4.34)<br />

where M is the length of the path and for 1 ≤ j ≤ M the values p j are given by<br />

p j = s(v m ) (4.35)<br />

where m is the <strong>in</strong>dex of the j-th node of the path <strong>in</strong>to the vertex set V .<br />

45

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!