Overview
The main goal for introducing Biometric technique is to provide our user a personalized experience and security for cargo deliveries. This subsystem contains two parts, facial and fingerprint recognition. For facial recognition, AuBi will be able to recognize a registered user and display different greeting messages and user profile based on the identity of the user. During the cargo delivery process, fingerprint recognition will be used to verify the user’s authentication for retrieval to ensure the safety of the cargo.
Current Research
The development of the Biometric system contains two parts. Part one is focusing on implementation of facial recognition and combining with the realization of personalized display with UI subsystem. Part two is to use fingerprint recognition for identification.
AuBi take advantage of the well developed facial_recognition library in Python that is based on dlib library in C++ to realize facial recognition function. This dlib library is dependent on ResNet-34, a 34 layer convolutional neural network. By applying the facial_recognition library, it is able to recognize people based on one pre-entered photo for training. The result will be a String of name and then pass to other modules. It is successful to connect the camera to the Raspberry Pi computer and obtain pictures. The running for facial recognition program is also successful and provided an over 90% accuracy based on the test samples.
The fingerprint recognition function is based on Adafruit_CircuitPython library provided by the Adafruit Industries. The fingerprint signal transmits to the computer through UART. Users can easily enroll their fingerprints in AuBi and is able to retrieve their cargos by putting their registered finger on the fingerprint sensor. Also, users do not need to be recognized in facial recognition module to use their fingerprints for cargo retrieving. The results come out of the fingerprint module is only connected with the cargo module.
Literature Review
Fingerprint Identification
Among the many biometric features, the fingerprint is considered one of the most practical ones. Fingerprint recognition requires a minimal e6ort from the user, does not capture other information than strictly necessary for the recognition process and provides relatively good performance. Another reason for the popularity of fingerprints is the relatively low price of fingerprint sensors, which enables easy integration into PC keyboards, smart cards and wireless hardware. And it can hold a great performance for children. With standard hardware and software, the comparison for performance of fingerprints identification between children and adults is comparable to those of adults vs adults: for a FAR of 0.01% (0.0001), we obtain a TAR of 97.5%, and a TAR of 97.9% for a FAR of 0.1% (comparable with the TAR obtained for adults only: 98.39%).
For a commercial fingerprint sensor, it promise a False Acceptance Rate: <0.001% (Security level 3) and False Reject Rate: <1.0% (Security level 3)
[1]A. M. Bazen and S. H. Gerez, “Fingerprint matching by thin-plate spline modelling of elastic deformations,” Pattern Recognition, vol. 36, no. 8, pp. 1859–1867, 2003.
[2]J. Preciozzi et al., “Fingerprint Biometrics From Newborn to Adult: A Study From a National Identity Database System,” in IEEE Transactions on Biometrics, Behavior, and Identity Science, vol. 2, no. 1, pp. 68-79, Jan. 2020, doi: 10.1109/TBIOM.2019.2962188.
Facial Recognition
Facial recognition is a well-established technology that can do real time face recognition combined with the technology of facial detection, a computer technology that determines the locations and sizes of human faces in arbitrary (digital) images.
Verification algorithms used to match subjects to clear reference images (like a passport photo or mugshot) can achieve accuracy scores as high as 99.97% on standard assessments like NIST’s Facial Recognition Vendor Test (FRVT).[2]
[1]M. S. Kalas, “Real Time Face Detection and Tracking Using OpenCV,” International Journal of Soft Computing and Artificial Intelligence, vol. 2, no. 1, pp. 41–44, 2014
[2] Comparing rank-1 FNIR at N=1.6M FVRT 2018 mugshot photos for 2020 Yitu-004 algorithm (0.0008) and 2014 NEC-30 algorithm (0.041). Source: Patrick Grother, Mei Ngan, and Kayee Hanaoka, “FRVT Part 2: Identification,” March 27, 2020, https://github.com/usnistgov/frvt/blob/nist-pages/reports/1N/frvt_1N_report_2020_03_27.pdf and Patrick Grother, Mei Ngan, and Kayee Hanaoka, “FRVT Part 2: Identification,” November 2018, https://nvlpubs.nist.gov/nistpubs/ir/2018/NIST.IR.8238.pdf.
Voice Identification
There are several advantages for voice identification.
First, speech is a natural signal to produce that is not considered threatening by users and in many applications, speech may be the main (or only, e.g., telephone transactions) modality, so users do not consider providing a speech sample for authentication as a separate or intrusive step. Second, it is convenient especially in those situations that involve telephones because the telephone system provides a ubiquitous, familiar network of sensors for obtaining and delivering the speech signal. However, these gains may not be applicable to our project because voice is not the way of input that we must use.
Also, based on our user case that the operation of the robot will take place in the hallway, there will be a considerable amount of environmental noises which will decrease the accuracy of the voice identification and recognition. According to [1], the EER will raise to about 10% on conversational speech tested on the Switchboard database.
[1]F. Bimbot, J.-F. Bonastre, C. Fredouille, G. Gravier, I. MagrinChagnolleau, S. Meignier, T. Merlin, J. Ortega-García, D. PetrovskaDelacrétaz, and D. A. Reynolds. “A tutorial on text-independent speaker verification.” EURASIP Journal on Applied Signal Processing, vol. 4, pp. 430-451, 2004.
[2]Kramberger, I., Grasic, M., & Rotovnik, T. (2011). Door phone embedded system for voice based user identification and verification platform. IEEE Transactions on Consumer Electronics, Consumer Electronics, IEEE Transactions on, IEEE Trans. Consumer Electron, 57(3). https://doi.org/10.1109/TCE.2011.6018876