Low Quality Fingerprint Image Segmentation Technology Research

With the rapid development of the social information security, biometrictechnology has been widely used because of its life unique, reliable and convenientfingerprint recognition technology become the key research topic in manytechnologies, and is widely used in banking, public security, customs and newlyreleased third-generation ID cards and other related field of security. However,fingerprint recognition images collected due to the environmental impact, such as wetfingers, peeling, oil, static occurs, there will be varying degrees of noise pollution, theimage recognition is greatly reduced, so that the fingerprint image processing toincrease a lot of difficulty. Fingerprint image segmentation is a key part of thefingerprint recognition technology, especially in the fingerprint image segmentation,If you can not be good for fingerprint image segmentation, extract the fingerprintimage area (containing part of the fingerprint texture), subsequent fingerprint imageprocessing, such as fingerprint enhancement, refinement and the match will beseverely disrupted, fingerprint identification system error rate will be significantlyincreased, which will be seriously hampered the development of fingerprintrecognition technology. Difficult for low-quality fingerprint image segmentation, thecombination of BP neural network fingerprint image segmentation method,combining artificial neural networks and fingerprint recognition technology, some ofthe new method is feasible. Overall,In fingerprint recognition technology to make amore in-depth study of the following aspects.(1) Understand the various types of artificial neural network, BP neural networkin-depth research and a lot of experiments to determine the relevant parameters of theBP network is relatively reasonable, and BP neural network, which must be improvedso that it can more effectively applied to the fingerprint image processing.(2) Calculation method of fingerprint image orientation field, and based on thelocal gradient image of the fingerprint orientation field calculation and simulation.(3) a lot of research on fingerprint image segmentation method, BP network combined with the fingerprint feature points fingerprint segmentation algorithm,experiments show that this algorithm is compared with many other methods bettersegmentation results.(4) In the fingerprint image segmentation process, some residual segmentationbased on the image processing method of the averaging filter, the effect is better thanthe opening in the morphological image processing operation, better repairfingerprint.