Thermal handprint analysis for forensic identification using Heat-Earth Mover's Distance

Abstract

Recently, handprint-based recognition system has been widely applied for security and surveillance purposes. The success of this technology has also demonstrated that handprint is a good approach to perform forensic identification. However, existing identification systems are nearly based on the handprints that could be easily prevented. In contrast to earlier works, we exploit the thermal handprint and introduce a novel distance metric for thermal handprint dissimilarity measure, called Heat-Earth Mover's Distance (HEMD). The HEMD is designed to classify heat-based handprints that can be obtained even when the subject wears a glove. HEMD can effectively recognize the subjects by computing the distance between point distributions of target and training handprints. Through a comprehensive study, our identification system demonstrates the performance even with the handprints obtained by the subject wearing a glove. With 20 subjects, our proposed system achieves an accuracy of 94.13%for regular handprints and 92.00% for handprints produced with latex gloves.