Researchers improve gait biometrics recognition rates for older people
A method of authenticating older individuals with biometric gait recognition by a team of researchers from China and Italy, who applied a gait template synthesis algorithm to correct for fluctuations in gait, Phys.org reports.
Intra-subject gait fluctuation is significantly higher in older people than younger ones, according to the report. Gait could otherwise be a suitable biometric for wearable device authentication due to its fraud resistance, passivity, and continuousness.
Professor Li Ye, Dr. Sun Fangmin, and Dr. Zang Weilin at the Chinese Academy of Science’s Shenzhen Institute of Advanced Technology (SIAT), working with colleagues from the University of Calabria, published research in Information Fusion to extend work previously published in IEEE Internet of Things Journal in 2018, to older individuals.
The researchers tested the recognition rate of gait biometrics as measured by wearable healthcare devices from a public dataset released by Osaka University in 2011. They used an arbitration-based score-level fusion algorithm, in which two matching algorithms performed initial matches, and any inconsistent judgements were decided by a third algorithm. This improved the recognition rate compared to the existing histogram similarity-based method, and reached 96.7 percent in the tests.
Gait recognition has yet to reach substantial deployment, but police in China were reported earlier this year to have tested the technology for public surveillance.