| With the improvement of mobile technology and smart phone, payment throughsmartphone has received unprecedented attentions. With the improvement ofcommunications and information technology, people are increasingly using mobilephones to store and transfer large amounts of personal information or privacy, and hopeit can ensure security of personal information.Fingerprint recognition technology is currently used in smart phone for mostmature identity authentication and payment transaction authentication. Fingerprintrecognition algorithm research has a very wide range of important significance basedon smart mobile. The result of fingerprint recognition, is not only depends on the levelof fingerprint technology, it also depends on the fingerprint identification algorithm.Preprocessing is an important part of fingerprint recognition. In this paper, we focuseson preprocessing study involving three major steps: image enhancement,binarizationand refinement.We will use the DB2and DB4fingerprint database of FVC2004(FingerprintVerification Competition2004,2004fingerprint recognition competitions) to performperformance tests. Through FVC2004performance test simulation and experimentalverification of fingerprints, we can make the following conclusion: Based on the samefeature extraction and matching algorithms, fingerprint image preprocessingalgorithm proposed by this paper is more efficient than the traditional algorithms inthe area of noise eliminating, avoid all kinds of fake feature information.The research done by this paper for solving the problems above are as follows:1. Fingerprint image enhancement algorithm based on Fourier and directionalfilterEnhancement algorithm based on Gabor filter performs well in dealing withhigh-quality fingerprint image, but the preprocessing speed is low. Enhancement algorithm based on Fourier filter’s speed is fast and performs well in dealing with lowquality fingerprint image. In this paper, we use directional filter and Fourier transformalgorithm for fingerprint image’s enhancement, and the experiment results show thatthis algorithm performs well in dealing with the fingerprint image2. Dynamic threshold binarization algorithm based on fixed thresholdBinarization algorithm based on fixed thresholds is simple, but will producepseudo feature information when processing low quality fingerprint image.binarization algorithm based on dynamic threshold is compensate for thedisadvantages of fixed-threshold algorithm, but it will also introduce noise.Dynamic threshold algorithm based on graph can achieve good noise immunityaccording to experiment in this paper.3. Based on improved OPTAthinning algorithmFast thinning algorithm use4areas pixel to judge fingerprint border points andremove the points gradually; OPTA algorithm use8removing templates and2reserving templates to determine whether the center pixel is removed or not.Improved OPTA thinning algorithm can reconstruct removing templates and reservingtemplates. this paper use square template to proceed images’ ridge tracing on thinnedfingerprint image based on improved OPTA thinning algorithm Experiment resultsshow that compared with the traditional method, algorithms proposed in this paperperform well in removing pseudo features and overcoming interference factors. |