Research Of Fingerprint Recognition And Applications Based Linux+ARM

Fingerprint recognition is the most widely used and development in the biometric technology. The core technology of Fingerprint recognition is Fingerprint preprocessing,This paper focuses on fingerprint segmentation and enhanced algorithm among the fingerprint image preprocessing algorithm.. By analyzing that the gradient threshold segmentation algorithm and Gabor wavelet enhancement algorithm exist deficiencies, proposed improved algorithm,through comparative analysis of the experiment,we can achieve better results. To enhance the security of ATM machines,designed a fingerprint prototype system with fingerprint recognition features and friendly interface for ATM’s based on ARM CORTEX A8 Processor. Specific tasks include:First of all.read the literature and analysis features of biometric technology, discovering the advantages of fingerprint technology. Analysis the research status and development trend of fingerprint recognition technology at the present stage. And proposed the significance of research that combined fingerprint technology with the practical application. Outline the mathematical models’s principles and physical models’s calculation methods of the fingerprint image field in detailed. According to the theory of image orientation field、gradient field and frequency field,the paper emphatically studies the fingerprint segmentation that based on differential box dimension combined with gradient threshold,and the fingerprint enhancement techniques that based on the image field analysis. Analysised the problems of algorithms, and proposed improvement program.Then,for the question of the clear fingerprint image places is blurred after the global smoothing, in order to achieve the purpose of removing the real noise and not reduce the quality of overall fingerprint image,proposed local image smoothing algorithm,only smoothed the protrusion of direction field. In the process of image segmentation, segmentation threshold is fixed,but the segmentation result in local areas is not idea. For this problem,select the appropriate segmentation threshold A by using the gray-scale difference histogram. Compared with the gradient value T,when T> A,it is foreground. Otherwise,it is background. Experimental results show that the algorithm’s fore/background segmentation effect is more pronounced.for the distortion of pixel position and grayscale and showed normal random distribution when capture fingerprint image in non-ideal environments,proposed that image can be made Gaussian smoothing,adjust the position relationships of the pixel and gray value,so that the image is more suitable for Gabor function enhancement.For problems that on the direction of the field where big changes are not effective in fingerprint image enhancement, proposed operation using Gabor templates, adjust the template size dependingon the value of direction field,reduce false feature points and feature points lost,thus the accuracy also improved.Finally,combined with the practical application of ATM systems,given its solutions, and to achieve a fingerprint input,fingerprint matching fingerprint system prototype. The system prototype hardware is built on FPM10A ingerprint recognition module,S5PV210 processor,and software is built on Linux Qt graphics interface.