Deep Learning Classification based on Merge Ear and Single Finger Geometry Dataset
Balkees Ahmed Mohammed, Ziad M. Abood · 2022
Since the development of technology is always going on, we needed more accurate and secure methods, so we needed to strengthen the fingerprints that a person must get rid of from continuous impersonations and because of the series of crimes, fingerprints and DNA have become one of the most common biometrics among people, but with Advances in internet technology and the interplay between computers and the things around us, other forms of biometrics must be sought for more security. For example, the distinctive physical characteristics of users in computer science are increasingly being used as forms of identification and access restriction. Mobile devices use fingerprints, eye scanning, and face recognition. Other biometrics use iris, veins, and palm prints. The ear is also a potential biometric method, and one-finger engineering is a new and powerful way to detect and identify people. 70% of the samples were trained and 30% of the samples were tested, and they were processed by using one of the primary processing methods, the Convolutional Neural Networks method, which eliminates the noise resulting from the imaging process because the data was built, and then the bitwise filter method was used, where eight layers were used. As a result of merging the images, the distinguishing characteristics were extracted, and then the person was classified. The study proved the success of the method, giving an accuracy rate of 99.8%.