User recognition using sequential footprints under shoes based on mat-type floor pressure sensor

A novel technique with sequential footprints measured by a mat-type floor pressure sensor is proposed to make a possible footprint-based solution for user recognition problem under shoes, specially focused on the problem of shoeprint-to-shoeprint comparison and the discrimination of different persons' footprints with similar shoes using shoeprint-to-barefoot footprint compar ison. The main idea for shoeprint-to-barefoot footprint comparison consists of three parts: To use not the footprint shape information, which is contaminated by the shape of the shoes and its sole structure, but the behavioral characteristics of human walking, sequential footprints are used as a special form of overlapped image, Footprint Energy Image (FEI). It is an image whose pixel value is determined by how long user keeps in contact with mat-type floor pressure sensor during a single gait cycle of walking. Second, to consider the difference of alignment between shoes and its barefeet inside the shoes, an iterative algorithm which can find the most proper rotation angle and center point for template matching is developed. And, to compare the dissimilarity between a given footprint under shoes and previously registered barefoot footprints of various users, a weighting window is proposed depending on the shape and its value of the reference barefoot FEI for the purpose of intensifying the center area of image comparison and less regarding the area near outlines. The effectiveness of the proposed method is verified by the successful identification of 24 university students' footprints.