In fact, the correct functioning of any fingerprint recognition system
depends directly or indirectly on the OF information. Recently,
different methods have been proposed for improving the accuracy
and speed of fingerprint identification, verification and classification,
in all of which extracting the OF of fingerprint images has been
as the basic procedure [6,7].
There are many reasons for extensive usage of Markov chains
in digital signal and image processing. According to the nonstationary
identity of fingerprint images, it seems to bemore useful
to analyze such 2-D digital signals in terms of Hidden Markov Models
[8–10]. Moreover, because of its regular texture pattern, the
fingerprint ridge OF can be viewed as a Markov chain mentioned
in [11]. The matching approach proposed in this paper uses an
improved HMM structure based on fingerprint OF around a reference
point. In the suggested HMM, the essential information for
the matching procedure has been extracted based on an ergodic
topology and the related training operations. The higher accuracy
and robustness of the proposed method has been proved by reliable
evaluation experiments.
The rest of this paper is organized as follows: In Section 2, a
general overview of HMM theories and applications is given. Various
HMM topology species with their differences are discussed
in the first subsection of Section 3. The proposed method advantages
are also listed in this subsection. Section 3 also comprises
the implementation steps for the proposed matching method. The
evaluation strategies accompanied with their related experimental
results are specifically stated in Section 4. Finally, Section 5
includes discussions and conclusions of the proposed matching
method.