We study the problem of score normalization in biometric verification systems. Specifically, we introduce a new class of normalization techniques, which unlike the commonly used parametric score normalization techniques, such as z- or t-norm, make no assumptions regarding the shape of the underlying score distribution. The proposed class of normalization techniques first estimates the relevant score distribution in an impostor-centric manner using kernel density estimation and then maps the estimated distribution to a common one. Our experimental results obtained on the FRGCv2 face database show that the proposed non-parametric score normalization techniques consistently outperform their parametric counterparts when the target distribution takes a log-normal form and that all assessed techniques, i.e., z-, t-, zt- and tz-norms, improve upon the setting where no score normalization is used. We demonstrate that the normalization contribute significantly to the robustness of the recognition system, when missmatched conditions caused, for example, by lighting or ageing effects, between probe and gallery images are present.
COBISS.SI-ID: 9520468
Face recognition in uncontrolled environments remains an open problem that has not been satisfactorily solved by existing recognition techniques. In this paper, we tackle this problem using a variant of the recently proposed Probabilistic Linear Discriminant Analysis (PLDA). We show that simplified versions of the PLDA model, which are regularly used in the field of speaker recognition, rely on certain assumptions that not only result in a simpler PLDA model, but also reduce the computational load of the technique and ‐ as indicated by our experimental assessments ‐ improve recognition performance. Moreover, we show that, contrary to the general belief that PLDA‐based methods produce well calibrated verification scores, score normalization techniques can still deliver significant performance gains, but only if nonparametric score normalization techniques are employed. Last but not least, we demonstrate the competitiveness of the simplified PLDA model for face recognition by comparing our results with the state‐of‐the‐art results from the literature obtained on the second version of the large‐scale Face Recognition Grand Challenge (FRGC) database.
COBISS.SI-ID: 9520724
This paper focuses on the use of Gaussian Mixture models (GMM) for 3D face verification. A special interest is taken in practical aspects of 3D face verification systems, where all steps of the verification procedure need to be automated and no meta-data, such as pre-annotated eye/nose/mouth positions, is available to the system. In such settings the performance of the verification system correlates heavily with the performance of the employed alignment (i.e. geometric normalization) procedure. We show that popular holistic as well as local recognition techniques, such as principal component analysis (PCA), or SIFT-based methods considerably deteriorate in their performance when an “imperfect” geometric normalization procedure is used to align the 3D face scans and that in these situations GMMs should be preferred. Moreover, several possibilities to improve the performance and robustness of the classical GMM framework are presented and evaluated: i) explicit inclusion of spatial information, during the GMM construction procedure, ii) implicit inclusion of spatial information during the GMM construction procedure and iii) on-line evaluation and possible rejection of local feature vectors based on their likelihood. We successfully demonstrate the feasibility of the proposed modifications on the Face Recognition Grand Challenge data set.
COBISS.SI-ID: 9519444
The existing face recognition technology has reached a performance level where it is possible to deploy it in various applications providing they are capable of ensuring controlled conditions for the image acquisition procedure. However, the technology still struggles with its recognition performance when deployed in uncontrolled and unconstrained conditions. In this paper, we present a novel approach to face recognition designed specifically for these challenging conditions. The proposed approach exploits information fusion to achieve robustness. In the first step, the approach crops the facial region from each input image in three different ways. It then maps each of the three crops into one of four color representations and finally extracts several feature types from each of the twelve facial representations. The described procedure results in a total of thirty facial representations that are combined at the matching score level using a fusion approach based on linear logistic regression (LLR) to arrive at a robust decision regarding the identity of the subject depicted in the input face image. The presented approach was enlisted as a representative of the University of Ljubljana and Alpineon d.o.o. to the 2013 face-recognition competition that was held in conjunction with the IAPR International Conference on Biometrics and achieved the best overall recognition results among all competition participants. Here, we describe the basic characteristics of the approach, elaborate on the results of the competition and, most importantly, present some interesting findings made during our development work that are also of relevance to the research community working in the field of face recognition.
COBISS.SI-ID: 10276180
The paper introduces a novel framework for 3D face recognition that capitalizes on region covariance descriptors and Gaussian mixture models. The framework presents an elegant and coherent way of combining multiple facial representations, while simultaneously examining all computed representations at various levels of locality. The framework first computes a number of region covariance matrices/descriptors from different sized regions of several image representations and then adopts the unscented transform to derive low-dimensional feature vectors from the computed descriptors. By doing so, it enables computations in the Euclidean space, and makes Gaussian mixture modeling feasible. In the last step a support vector machine classification scheme is used to make a decision regarding the identity of the modeled input 3D face image. The proposed framework exhibits several desirable characteristics, such as an inherent mechanism for data fusion/integration (through the region covariance matrices), the ability to examine the facial images at different levels of locality, and the ability to integrate domain-specific prior knowledge into the modeling procedure. We assess the feasibility of the proposed framework on the Face Recognition Grand Challenge version 2 (FRGCv2) database with highly encouraging results. Note that the work presented in this paper could easily be applied to other biometric (image) modalities, such as 2d face images, as well, as depth images can be considered as a form of intensity images.
COBISS.SI-ID: 9821012