In the context of deep learning's application in biometric recognition, the challenge faced is the lack of a sufficiently large training sample available for a particular biometric modality. The extra equipment also makes the recognition expensive and difficult to scale. One disadvantage of biometric systems is the need for extra equipment, such as high-resolution cameras required for iris recognition and fingerprint readers for fingerprint recognition. While biometric recognition is being increasingly employed for the sake of the security perspective, there remain challenges that still need to be addressed for the next step in the integration of biometrics into more institutions. Deepak Kumar Sharma, in Trends in Deep Learning Methodologies, 2021 2.7 Challenges in biometric recognition and security These include iris-scanning and foot-scanning technology, and voice pattern ID, as well as facial recognition technologies ( Wolf, 2012).Īkash Dhiman. In early August 2012, Michael Bloomberg, Mayor of New York, and Ray Kelly (NYPD Commissioner) unveiled a new police surveillance infrastructure developed by Microsoft called the Domain Awareness System, which links existing police databases with live video feeds from a variety of different sources.įurthermore, according to a Homeland Security newswire in the United States, billions of dollars are being invested in the development of various biometric technologies capable of identifying anyone anywhere in the world. Furthermore routine phishing activities through the Internet and social networks provide a fertile ground for not only a simple one-dimensional set of data (photos and other personal data) but potentially three-dimensional datasets of associated friends, links, habits, and quite often current location. It is not surprising to note that to put such a database together security agencies never (at least we cannot identify any evidence) consult or seek permission to keep people's records in their data centers. For the system to be effective, the matching database should be as wide and comprehensive as possible. CCTVs in streets, public places, and office buildings record images 24/7, sophisticated algorithms then carry out a matching exercise with an existing database of images of potential “villains” or “targets.” A match will trigger enhanced surveillance and possible future and further action. The way the technology works is straightforward. Following the 9/11 terrorist attack, the use of facial recognition, especially in crowded places, as a means of detecting possible threats has been debated widely. With the ever-growing number of video surveillance cameras mushrooming in large cities, the use of the data captured by these cameras has been at the center of a number of privacy and human rights storms.
Fingerprint recognition is one of the first and original biometric technologies that have been grouped loosely under digital forensics. Steve Wright, in Strategic Intelligence Management, 2013 Biometric Technologiesīiometric technologies generally refer to the use of technology to identify a person based on some aspect of their biology. 2 shows the production process of two different collaborative ways to imitate fake fingerprints, and the left side of the flowchart is a collaborative approach. Putte reported that six of the fingerprint sensor spoofing incidents were caused by a collaborative approach, in which the artificial fake fingerprint was dummy silicon rubber. In cooperation, artificial fingerprints are created directly from alive fingerprints with person's consent using casts and molds while in non-cooperative method, artificial fingerprints are created indirectly from latent or cadaver fingerprints on objects or bodies.
Putte in 2000, divided the artificial fingerprint replicas into two categories: cooperation of fingerprint owner and non-cooperative of fingerprint owner. Former research pointed out that fingerprint sensors are easily spoofed by some artificial replicas using wax, moldable plastic, Play-Doh, clay or gelatin.
Although fingerprint identifier verification has become a popular and efficient security technology, most of fingerprint recognition systems are vulnerable to various threats, such as attacks at the sensor level or at the database level.