Research On Fingerprint Enhancement And Matching Algorithm Under Embed System

With the maturation of fingerprint recognition technology and related equipment supporting, especially the improvement of the fingerprint sensor and the chips, embedded fingerprint recognition systems are increasingly being seen as acceptable. Market demand is also growing. Embedded Fingerprint Identification System also becomes a meaningful and practical application of the research topic. Now, quite a number of domestic and foreign scholars and enterprises made a lot of research on on-line fingerprint identification system, on-line fingerprint recognition technology have matured, but the construction of embedded fingerprint identification system is relatively backward. Relative to the on-line fingerprint identification system, embedded fingerprint identification system face some difficult:First, the limited space in the embedded environment, the second is the relatively slow speed of the chip operation, the third is embedded environment storage space is relatively small.In this paper, the above three issues in-depth study, the main research contents include:Facing the issue of the limitation of storage space in the embedded environment, the fingerprint is processed line by line. All processing steps are based on the block. Using this strategy the space consumption is reduced.Variable template enhancement algorithm is proposed. One-dimensional or two-dimensional templates are used in most of the existing enhancement algorithm. Due to the more using of the neighborhood pixels, two-dimensional template can achieve more acceptable enhancement results. But drawback is more time consuming. One-dimensional templates using fewer pixels to enhance the fingerprint speed up the enhancement, but the loss of performance. In this paper, one-dimensional variable template is introduced, according to the specific circumstances of neighborhood pixel the shape and size of the template is dynamically changed. This strategy of template selection can highlights the fingerprint ridge structure more accurately, while speed up the enhancement speed. Three-valued feature vectors based fingerprint matching algorithm is proposed. Many of the existing fingerprint matching method using certain aspects of neighborhood characteristics as character descriptions. First, the characteristics of a single context can not fully describe the details of the neighborhood information, but additional feature information will make the character vector longer, increasing the storage space and reading time. In our algorithm, the direction of minutiae neighborhood field, neighborhood minutiae, neighborhood ridges are used as descriptions. and all descriptions are stored as three-valued feature vectors. Three-valued feature vectors based fingerprint matching algorithm not only improves the speed of the fingerprint match, but also reduce the storage space. Store more fingerprints to improve the system capacity of fingerprint.