Research And Implementation Of Fingerprint Recognition Algorithm Based On ARM9 Embedded System

Fingerprint recognition technology is a biometric technology. Because the fingerprint is unique, permanent, universal and acquisition of available, technology of fingerprint identifi- cation is widely used. At present the majority of high-performance automated fingerprint identification system is based on the PC platform, and then, there are disadvantages, bulky volume, difficulty of carring, time performance, higher price,and so on. The superiorities of embedded devices are small volume, real-time high, low prices. Thus fingerprint identification has broad application prospects in embedded devices.Firstly, this paper describes development overviewthe of fingerprint recognition technology and embedded technology. Secondly, we analyze algorithms of fingerprint image process and fingerprint feature extraction that have been proposed at home and abroad. Combining the needs of practical application, we realize the image preprocessing by three steps: directional image smoothing, image binarization and image thinning, and get binary thinned image. Determining the location of the core point by using the principle of support vector machine, we propose the algorithm that is the core point of the fingerprint to establish the details of a cartesian coordinate system fingerprint feature point extraction. We increase the link of image quality assessment to improve the matched reliability before the fingerprint matching. Based on existent triangle similar algorithm used in fingerprint matching, we use the algorithm to match fingerprint ,that is similar triangles based on the core points of the fingerprint matching algorithm and can reduce the time and space complexity, which will help in embedded to achieve. Finally, the paper simulates embedded fingerprint identification system in PC and ARM9 , analyses false acceptance rate (FAR), false rejection rate (FRR) and real-time. Experiments show that the fingerprint identification process, based on the core point of the embedded fingerprint recognition algorithm to match the fingerprint image from the collection point and the size effect, there is a better recognition effect.