December 15, 2014 -
A student research team at Cal State Fullerton is currently working on a new process to improve personal security using multimodal biometric authentication, according to a report by Daily Titan.
The CSUF research team consists of seven students led by computer science professor Mikhail Gofman and information systems and decision sciences professor Sinjini Mitra, both of whom are experts in biometrics fusion research and biometric authentication, respectively.
The team recently presented at the CSUF Southern California Conference for Undergraduate Research.
Their focus is on improving smartphone and tablet security by developing a multimodal framework, which uses multiple inputs, such as face and voice, to unlock or secure these devices.
The research team believes that improved biometric authentications could provide more secure authentication protocols than traditional systems that use passwords and pins.
Additionally, the team’s research also concluded that current sensor technology and biometric algorithms are not strong enough to stand up against advanced spoofing attacks or tend to not work well in less than ideal conditions.
The team tested both Fischer Faces and Hidden Markov Model algorithms, which are typically used in face and voice recognition.
Based on their tests, the algorithms were found to have 40% error rates. But when the team experimented using their own approach they were able to cut the error rates down to nearly 20%.
“Our preliminary results indicate that such ‘multimodal’ fusion yields significantly more accurate authentication results than methods that use only one single biometric trait,” Mitra said.
The team’s prototype allows the user to record a brief video of their face while speaking a sentence. The prototype then extracts the photographs and audio from the video, and statistically combines the samples to generate a third level of integrated biometrical inputs.
In the future, the team also plans to include fingerprinting in the process because many smartphones are now equipped with built-in fingerprint scanners.
To date, the research team has only experimented on the Android platform, but it hopes to soon conduct research on Apple’s iOS.