NASK - Sense of telecommunications HandTherm - Research Projects -

The project consisted in the implementation and assessment of the effectiveness of
a biometric identification method based on thermal hand images. This mode of biometric verification offers numerous benefits, including quick, non-invasive and hygienic measurements and non-reliance on external lighting sources, which is an issue with photographs taken in visible light. In addition, heat distribution is a modality which is inaccessible to bystanders, and forging a heat signature would pose a challenge for criminals. The flaws of this method include changing heat patterns due to external conditions and human physiology (health, physiological and metabolic processes), which is why it is important to identify information which does not change with time. The project involved the implementation of a feature extraction and classification method based on parametric texture descriptors (LBP, BSIF, Gabor wavelets) and recently-popular convolutional neural networks (CNN). Preliminary research demonstrated that heat distribution patterns are individualised and can be used to achieve accuracy exceeding 90%. More information can be found in the article “Unconstrained Biometric Recognition based on Thermal Hand Images” presented at the IWBF 2018 international conference.