A Criterion for Analysis of Different Sensor Combinations with an Application to Face Biometrics

Abstract

In this paper, we propose a criterion for pairwise combination of information from different sensors in order to decide how a given pair of sensors is useful for different applications. This criterion is related to the principle of maximum information preservation. We present experimental results for the case of face images at different spectral bands, which allow for the in advance evaluation of the usefulness of different sensor combinations as well as the possibility for crossed-sensor recognition (matching of images acquired in different spectral bands). The criterion that we propose is a generalization of the Fisher score for the case of mutual information, which is measured as the ratio of the interclass information to the intraclass. The score we propose measures the behavior of a pair of sensors either when they are used in combination or when they are used to discriminate between classes. Based on Information Theory measurements, we conclude that the best spectral band combination always contains the thermal image, while the best combination for crossed-sensor recognition is VIS and NIR.

Performance estimation of the state-of-the-art convolution neural networks for thermal images-based gender classification system

“…All the trained models are tested on the combination of the two different datasets including Carl 13,6 and the locally gathered indoor thermal dataset. This is done to cross-validate the effectiveness of all the trained classifiers, as discussed in Sec.…”

“…All these studies measured differences between the left and right side of different areas of the head. 6,7,8 The literature reports that in healthy subjects the difference in skin temperature from side to side of the human body is as small as 0.2°C. 8 The heat emission from the human body is related to cutaneous vascular activity, yielding enhanced heat output on vasodilation, and reduced heat amount on vasoconstriction.…”

“…In this study, the Tufts thermal faces [10][11][12] and Carl thermal faces datasets 13,6 are used to train and test a selection of state-of-the-art neural networks to perform the gender classification task. Figure 1 shows some examples of thermal facial images with varying poses from the Tufts dataset and frontal facial poses from the Carl dataset.…”

Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models

“…1 the first step includes data collection and data acquiring using a thermal camera. In this study, we have used three publically available thermal face datasets which include tufts [22][23][24], carl [27][28], and laval facial thermal dataset [29]. Tufts dataset [22][23][24] has been published recently with data samples from 6 different image modalities which include visible, near-infrared, thermal, computerized sketch, a recorded video, and 3D images.…”

A Comprehensive Database for Benchmarking Imaging Systems

“…Other thermal face images are Carl database [124], [125] Terravic Facial IR Database [112] and Kotani Thermal Facial Emotion (KTFE) [126]. However, availability of thermal image face databases is limited: there is a lower demand for thermal images as compared to 2d databases, and thermal images are significantly more difficult to obtain.…”