Published Paper Details:

Now a day almost each and every application is secured with some authentication technique and one among them is finger print authentication which is present in each and every mobile devices. This fingerprint recognition and authentication comes under the field of pattern recognition. This pattern recognition is example of mobile devices which use sensors for authenticating the users. As we all know that in some situations these finger print authentication is not giving accurate results when compared with contact based to contact less. Hence this motivated me to design an application using CNN(Convolutional Neural Networks) for fingerprint recognition. By using CNN model, we can design the system very accurately for identifying and match both contactless and contact-based fingerprint images. In order to design this model we initially collect finger print dataset from NIST dataset, which contains more than 2400 finger prints, which paired into 1200 paired fingerprints. In this we try to collect both identical and non-identical images and then try to train the system with these images. Once the system is trained with these images we can able to test some sample images and find whether the CNN model is accurately identifying the images either contact based or contact less. Our experimental results clearly state that our method can give accuracy of more than 90 percent compared with primitive ML models.