Master Thesis - Department of Computer Science
Master Thesis - Department of Computer Science
Master Thesis - Department of Computer Science
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Figure 2.6: (a) and (c) are Input Images; (b) and (d) are enhanced recoverable regions<br />
superimposed on corresponding input images.<br />
have optimal joint resolution in both spatial and frequency domains [32]. Examples<br />
<strong>of</strong> fingerprint enhancement with Gabor filtering as proposed by Hong, Wan , and Jain<br />
are shown in Fig 2.6.<br />
2.2.2 Feature Representation and Feature Extraction<br />
A good fingerprint representation should have the following two properties: saliency<br />
and suitability. Saliency means that a representation should contain distinctive infor-<br />
mation and suitability means that the representation can be easily extracted, stored<br />
in a compact fashion, and be useful for matching. The fingerprint pattern, when<br />
analyzed at different scales, exhibit different features.<br />
• At global level: The ridge line flow, singular points (loop and delta), external<br />
fingerprint shape, orientation image, frequency image.<br />
• At local level: Minutiae (ridge termination and ridge bifurcation).<br />
• At very-thin level: Intra-ridge details (sweat pores).<br />
An elegant and practical method based on the Poincare index was proposed by<br />
Kawagoe and Tojo [58]. Let G be a vector field and C be a curve immersed in G.<br />
28