Gabriel Hermosilla

Fusion of Visible and Thermal Descriptors Using Genetic Algorithms for Face Recognition Systems

The aim of this article is to present a new face recognition system based on the fusion of visible and thermal features obtained from the most current local matching descriptors by maximizing face recognition rates through the use of genetic algorithms. The article considers a comparison of the performance of the proposed fusion methodology against five current face recognition methods and classic fusion techniques used commonly in the literature. These were selected by considering their performance in face recognition. The five local matching methods and the proposed fusion methodology are evaluated using the standard visible/thermal database, the Equinox database, along with a new database, the PUCV-VTF, designed for visible-thermal studies in face recognition and described for the first time in this work. The latter is created considering visible and thermal image sensors with different real-world conditions, such as variations in illumination, facial expression, pose, occlusion, etc. The main conclusions of this article are that two variants of the proposed fusion methodology surpass current face recognition methods and the classic fusion techniques reported in the literature, attaining recognition rates of over 97% and 99% for the Equinox and PUCV-VTF databases, respectively. The fusion methodology is very robust to illumination and expression changes, as it combines thermal and visible information efficiently by using genetic algorithms, thus allowing it to choose optimal face areas where one spectrum is more representative than the other.

An enhanced representation of thermal faces for improving local appearance-based face recognition

AbstractThis paper proposes a new methodology to improve appearance-based thermal face recognition methods by using an enhanced representation of the thermal face information. This new representation is obtained by combining the pixels of the thermal face image and the vascular network information that is extracted from the same thermal face image. The effect of using the enhanced representation is evaluated for 5 different face recognition methods (LBP, WLD, GABOR, SIFT, SURF) in two public thermal face databases (Equinox and UCHThermalFace). The experimental results show that the proposed enhanced representation improves the performance of most of the analyzed appearance-based methods. The largest improvements are obtained when this representation is used together with methods based on the Gabor Jet Descriptor (GJD), the Weber Linear Discriminant (WLD) and Speeded Up Robust Features (SURF). In general terms the improvement is larger in indoor setups than in outdoors.

Study of Local Matching-Based Facial Recognition Methods Using Thermal Infrared Imagery

This paper shows a comparative study among different local matching-based methods for thermal infrared face recognition. The principal assumption of this work is that the thermal face corresponds to the diffuse energy emission captured by an infrared camera, where the thermal signature is unique for each subject and it can be addressed as a texture descriptor with thermal images. Local matching-based methods find inter-class differences that improve the face recognition rate in thermal spectrum. Specifically, this work considers four methods: Local Binary Pattern (LBP), Local Derivative Pattern (LDP), Weber Linear Descriptor (WLD) and Histograms of Oriented Gradients Descriptors (HOG). The methods are evaluated and compared using the UCHThermalFace database, that considers real-world conditions and unconstrained environments, such as indoor and outdoor setups, natural variations in illumination, facial expression, pose, accessories, occlusions, and background. Results indicate that HOG variants followed by LBP method achieved the best recognition rates for face recognition systems.

Thermal Face Recognition Using Local Patterns

The aim of this article is to compare the performance of well-known visible recognition methods but using the thermal spectrum. Specifically, the work considers two local-matching based methods for face recognition commonly used in visible spectrum: Local Binary Pattern (LBP) and Local Derivative Pattern (LDP). The methods are evaluated and compared using the UCHThermalFace database, which includes evaluation methodology that considers real-world conditions. The comparative study results shown that, contrary to what happens in the visible spectrum, the LBP method obtains the best results from the thermal face recognition. On the other hand, LDP results show that it is not an appropriate descriptor for face recognition systems in the thermal spectrum.

Thermal Face Recognition in Unconstrained Environments Using Histograms of LBP Features

Several studies have shown that the use of thermal images can solve limitations of visible spectrum based face recognition methods operating in unconstrained environments. The recognition of faces in the thermal domain can be tackled using the histograms of Local Binary Pattern (LBP) features method. The aim of this work is to analyze the advantages and limitations of this method by means of a comparative study against other methods. The analyzed methods were selected by considering their performance in former comparative studies, in addition to being real-time—10 fps or more—to require just one image per person, and to being fully online (no requirements of offline enrollment). Thus, in the analysis the following local-matching based methods are considered: Gabor Jet Descriptors (GJD), Weber Linear Discriminant (WLD) and Local Binary Pattern (LBP). The methods are compared using the UCHThermalFace database. The use of this database allows evaluating the methods in real-world conditions that include natural variations in illumination, indoor/outdoor setup, facial expression, pose, accessories, occlusions, and background. In addition, the fusion of some variants of the methods was evaluated. The main conclusions of the comparative study are: (i) All analyzed methods perform very well under the conditions in which they were evaluated, except for the case of GJD that has low performance in outdoor setups; (ii) the best tradeoff between high recognition rate and fast processing speed is obtained by LBP-based methods; and (iii) fusing some methods or their variants improve the results up to 5 %.

A Neural Network Approach for Building An Obstacle Detection Model by Fusion of Proximity Sensors Data

Proximity sensors are broadly used in mobile robots for obstacle detection. The traditional calibration process of this kind of sensor could be a time-consuming task because it is usually done by identification in a manual and repetitive way. The resulting obstacles detection models are usually nonlinear functions that can be different for each proximity sensor attached to the robot. In addition, the model is highly dependent on the type of sensor (e.g., ultrasonic or infrared), on changes in light intensity, and on the properties of the obstacle such as shape, colour, and surface texture, among others. That is why in some situations it could be useful to gather all the measurements provided by different kinds of sensor in order to build a unique model that estimates the distances to the obstacles around the robot. This paper presents a novel approach to get an obstacles detection model based on the fusion of sensors data and automatic calibration by using artificial neural networks.

Easy-to-Build Textile Pressure Sensor

This article presents the design, construction, and evaluation of an easy-to-build textile pressure resistive sensor created from low-cost conventional anti-static sheets and conductive woven fabrics. The sensor can be built quickly using standard household tools, and its thinness makes it especially suitable for wearable applications. Five sensors constructed under such conditions were evaluated, presenting a stable and linear characteristic in the range 1 to 70 kPa. The linear response was modeled and fitted for each sensor individually for comparison purposes, confirming a low variability due to the simple manufacturing process. Besides, the recovery times of the sensors were measured for pressures in the linear range, observing, for example, an average time of 1 s between the moment in which a pressure of 8 kPa was no longer applied, and the resistance variation at the 90% of its nominal value. Finally, we evaluated the proposed sensor design on a classroom application consisting of a smart glove that measured the pressure applied by each finger. From the evaluated characteristics, we concluded that the proposed design is suitable for didactic, healthcare and lifestyle applications in which the sensing of pressure variations, e.g., for activity assessment, is more valuable than accurate pressure sensing.

Face Recognition and Drunk Classification Using Infrared Face Images

The aim of this study is to propose a system that is capable of recognising the identity of a person, indicating whether the person is drunk using only information extracted from thermal face images. The proposed system is divided into two stages, face recognition and classification. In the face recognition stage, test images are recognised using robust face recognition algorithms: Weber local descriptor (WLD) and local binary pattern (LBP). The classification stage uses Fisher linear discriminant to reduce the dimensionality of the features, and those features are classified using a classifier based on a Gaussian mixture model, creating a classification space for each person, extending the state-of-the-art concept of a “DrunkSpace Classifier.” The system was validated using a new drunk person database, which was specially designed for this work. The main results show that the performance of the face recognition stage was 100% with both algorithms, while the drunk identification saw a performance of 86.96%, which is a very promising result considering 46 individuals for our database in comparison with others that can be found in the literature.