Deep ContactlessS 3D-Middle Finger Knuckle Recognition

Abstract

Automation security is one of the main concerns of modern times. A safe and reliable identity verification system is in great demand. A biometric verification system can represent a reliable method of identifying an individual. The knuckle pattern is considered to be one of the emerging hand biometrics because it has the potential to identify individuals. This presented work explores the possibility of using 3D middle finger knuckle for biometric identification. The study provides a new simple, trained from scratch but effective deep convolutional neural network model that designed for 3D figure knuckle print recognition. To the best of the author's knowledge, this is the first successful attempt to develop 3D middle finger knuckle based authentication system using Convolutional Neural Network (CNN). Extensive experiment was carried out using (HKPolyU) the 3D knuckle image database of Hong Kong Polytechnic University. The performance of the proposed CNN model has been evaluated using two session data obtained from different camera lenses. The model designed to be implemented in a real-time system. It can be deployed in a small-scale environment like offices, houses, or personal devices, where the training is very easy. The experimental results were very encouraging and showed the potential of using the 3D middle finger knuckle pattern for biometric applications. The results confirmed that despite various challenges compared with other studies in the same field, the proposed method provides an optimistic solution with an accuracy of 71%.

CNN-Based Transfer Learning for 3D Knuckle Recognition

“…Moreover, the lack of studies about this type of biometric identifer has confrmed the importance of utilizing such CNN models. Since [10,12] are the only two previous studies that are available on the CNN-based method for the 3D fnger knuckle recognition issue. Besides, [12] represents the merely preceding paper that used the CNN-based approach for 3D middle fnger knuckle recognition.…”

“…Furthermore, the authors in [12] presented work that explores the potential for biometric identifcation utilizing 3D middle fnger knuckle patterns. Te research ofers a recent, simple, trained from scratch, but efective deep convolutional neural network model designed for 3D fnger knuckle pattern recognition.…”

“…For every 3D image, there are seven photometric stereo images. As a result, each subject in each session has access to forty-two (3D) forefnger photometric stereo images and forty-two (3D) middle fnger photometric stereo images [1,12].…”

“…Te author of [12] mitigates the impact of the issue with the middle fnger by proposing a CNN model to be used for testing middle fngers. Terefore, only the middle fnger has been used for the study.…”

“…Te selection of the subject was not correlated to any criteria. However, using more subjects is feasible while taking longer training and testing times [12].…”