Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab
Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab
Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
4.2 <strong>Classification</strong> based on features<br />
Energy<br />
∑w−1<br />
∑w−1<br />
f i = C i (x, y) 2 (4.16)<br />
y=0 x=0<br />
Entropy<br />
∑w−1<br />
∑w−1<br />
f i = C i (x, y) log ∗ (C i (x, y)) (4.17)<br />
y=0 x=0<br />
Inverse difference moment<br />
∑w−1<br />
∑w−1<br />
f i =<br />
y=0 x=0<br />
C i (x, y)<br />
1 + (x − y) 2 (4.18)<br />
Cluster shade<br />
∑w−1<br />
∑w−1<br />
f i = (x − m x + y − m y ) 3 C i (x, y) (4.19)<br />
y=0 x=0<br />
with<br />
∑w−1<br />
∑w−1<br />
P x (i) = C(i, j) P y (j) = C(i, j) (4.20)<br />
j=0<br />
i=0<br />
and<br />
∑w−1<br />
∑w−1<br />
m x = iP x (i) m y = jP y (j) (4.21)<br />
i=0<br />
j=0<br />
Cluster prom<strong>in</strong>ence<br />
∑w−1<br />
∑w−1<br />
f i = (x − m x + y − m y ) 4 C i (x, y) (4.22)<br />
y=0 x=0<br />
When all feature vectors F i for all images <strong>in</strong> I T have been calculated, an arbitrary image I α<br />
can be classified by us<strong>in</strong>g a classifier such as k-NN, LDA, CART or support vector mach<strong>in</strong>es<br />
(SVM). Throughout this thesis however the only classifiers used are k-NN and SVM. The<br />
classification process based on the features just calculated and the accord<strong>in</strong>g classifiers are<br />
described <strong>in</strong> section 4.4.<br />
41