The Fingerprint Image Classification And Matching Algorithm Research And Realization

Fingerprint-based identification is one of the most mature and dependable techniques, which refers to identifying an individual based on his or her physiological characteristics.Although there have been some AFIS (Automatic Fingerprint Identification System) in the market, the performance of these systems need to be improved further according to some peculiar application. In this paper, we make a deep research on several key algorithms:1. We proposed a simple, effective and adaptive segmentation method based on row projection histogram and the column projection histogram of the block standard deviation.2. We use the block orientation and poincare index to confirm the singular area, then classify the fingerprint effectively.3. We propose a minutiae matching method based on a new vector (line segment vector), using this vector, we can determine a truer reference point more efficiently, in addition, this new minutiae vector can also increase the accuracy of the minutiae matching.4. In order to transplant the algorithm to the embedded system, we also improve the algorithm of image enhancement and detection of minutiae.We develop the whole software system by VC++ 6.0, and the experiments are conducted on a standard fingerprint database fvc2004, the results show that our algorithm can get a improved performance.