Footprint Biometrics on Person Identification Using CNN Algorithm.

Footprint biometrics is as unique as a fingerprint in every human being. The prints on the sole of our foot can be used to identify individuals who have registered their information in our system. Biometric systems can be used as an alternative to password-protected profiles. With footprint as another option to strengthen a user's profile, the footprint biometric system can be used to secure confidential information of individuals. This study focuses on using Convolutional Neural Network Algorithm to acquire and identify footprints. The study uses the LeNet-5 architecture of CNN for data training and validation of the footprints registered. The system is composed of three major parts which are the platform where the user will step on to register their footprints and for the identification process as well. Raspberry Pi 3 Model B is used as the system's server where all necessary components are connected, and the UI of the system can be accessed in a computer. Raw registered images will undergo pre- processing where multiple filters will be applied before normalization such as Grayscale Filter, Anisotropic Diffusion Filter, and Gaussian Filter. 60 trials were done involving 10 test subjects in registering and identifying their footprints. The test yielded an accuracy of 91.67% considering the 5 misidentified trials.