Human body bioacoustics deliver 97 percent biometric accuracy, researchers say
Researchers from the Electronics and Telecommunications Research Institute (ETRI) in South Korea claim to have figured out how to accurately use bioacoustic body signatures to reveal identity, writes Digital Trends.
Published in the journal IEEE Transactions on Cybernetics, their research into next-generation and experimental biometrics argues sound waves passing through the body can be used to identify a person with 97 percent accuracy because they analyze properties like the individual’s skin, tension in joints and bone density.
“We can think of our body as a musical instrument that has a unique shape and composition of materials,” Joo Yong, one of the ETRI researchers on the project, told Digital Trends. “Our technology evaluates these traits of our body by vibrating a certain body part — for example, a hand — and hear[ing] the propagating sound as we alter the frequency of excitation. Our system is sophisticated enough to extract the features of our body as a proxy of a user’s anatomical and biomaterial properties and differentiate the individual with high accuracy.”
Biometric technology is susceptible to spoofing attacks because it currently leverages physiological characteristics to create images for identification, yet things could be completely different if human body vibration was used instead, the researchers say, because the unique spectral traits collected ca not be stolen, and could phase out spoofing to boost security. Bioacoustics spectral patterns, they found, do not suffer significant changes over time, but to ensure accuracy the researchers are looking into adding sensors to further integrate the technology with smartphones and wearables.
“In current biometrics — such as fingerprint, iris, and face recognition — one can make fake copies for spoofing because they rely on the structural features of the acquired image, and therefore once a template is stolen it can be a permanent threat to spoofing,” Yong said. “Our method uses characteristics inside the body and extracts information in the frequency domain, which creates a new level of security and makes it useful for applications that require a high level of security.”
Other potential next-generation biometrics include ear canal geometry recognition.