A New Biometric Identification System Based on Plantar Pressure

Abstract

Biometric recognition based on physiological characteristics has been widely used for security reasons. In these systems, unique features such as fingerprints, iris patterns, face structure, and voice are used to recognize a person instead of codes that can be easily transferred. The biometric identification system based on plantar pressure proposed in this work applies a support vector machine procedure to classify footprint patterns detected by a platform instrumented with fiber Bragg gratings. Two sets of 7 in-series gratings, with Bragg wavelengths within 1526.96 nm and 1553.86 nm range, delimit two sensing regions corresponding to the left and right footprints. The coupled and non-linear responses provided by the sensors are pre-processed and organized in a matrix. An algorithm based on support vector machine was used to extract and recognize foot features. The pattern recognition system constructed in this way returned hit rates greater than 86% for the 12 classes, corresponding to different stepping patterns, with data not belonging to the training set.