Fig 4 - uploaded by Matti Pietikäinen
Content may be subject to copyright.
Source publication
Recent developments in face analysis showed that local binary patterns (LBP) provide excellent results in representing faces.
LBP is by definition a purely gray-scale invariant texture operator, codifying only the facial patterns while ignoring the
magnitude of gray level differences (i.e. contrast). However, pattern information is independent of t...
Context in source publication
Context 1
... first normalized the training face samples to 24×24 pixels using the eye co- ordinates that are supplied with the datasets. Figure 4 illustrates some examples from the training samples. Our proposed facial representation involves the following free parameters to be fixed: the number and size of blocks when dividing the face images, the radius and number of neighbors for the LBP operator, the radius and number of neighbors for the contrast operator (VAR) and the quantization level. ...
Similar publications
This study examined the potential of integrating fractal texture with spectral information on urban landscape characterization by the maximum likelihood image classifier. The fractal texture was first derived from the red band of a Landsat ETM+ image by applying the triangular prism algorithm at different window sizes. The quality of the twenty-fiv...
Citations
... In another study authors combined LPB with a local contrast measure for face recognition. The authors also implicitly recognized that LBP cannot be considered as a contrast measure [107]. ...
... These include, for example, fields of target detection [181], traffic road visibility [182], [183], face recognition (through LBP) [106], image classification [184] and other computer vision applications. In a method combining LPB descriptors with a local contrast measure for face recognition has been proposed in [107]. In this method the local variance of pixel values is considered as a local contrast measure. ...
Contrast in visible images is one of the most relevant characteristics of visual signals. Since the pioneering works performed in vision psychology and optics, different definitions have been proposed in the literature. However, for the time being there exist no definition of contrast on which the vision research and visual processing scientific community can agree on. This makes it critical to have a clear view on the notion of contrast and its use in various applications. One issue to consider is how to define and particularly use this important measure in developing image processing and analysis methods. In this paper, we present a critical review of contrast measures and associated models developed by the scientific community in vision, optics, and image processing. We also provide learned lessons and guidelines on the use of appropriate contrast measures in selected visual information processing and analysis applications. We discuss challenges and propose research avenues for models enriched by recent findings in the field of human vision research and machine learning. We believe that this work serves as a guideline and can potentially open new research perspectives to the scientific community working on visual information processing and analysis.
... The rotation equivalence class considers the global rotation among LBP values of adjacent pixels and improves rotation invariance. Ylioinas et al. [12] , proposed a new method for gender recognition combining LBP features along with contrast information. The local contrast information (VAR) is calculated in a circular neighbour set. ...
... Composition of the best results from Best M-F parts of Tables 8 ,9 ,10 ,11 ,12 ,13 . The meanings of bold, underlined and shaded indicators are the same with the Table 8 . ...
Local binary patterns (LBP) are considered to be one of the most computationally efficient descriptor that can also be combined jointly among different variants to increase accuracy. In this study, we propose a method to obtain more discriminative 2D LBP features by optimizing projections of a joint LBP distribution onto the marginal histograms. To find a more efficient representation of the feature vector, we seek the least redundant marginal histograms of a joint LBP distribution via optimizing several constraints. In this way, we aim to have a more compact yet accurate feature vector in contrast to the methods that flatten the joint distribution. Experiments we perform on five popular texture datasets show that the feature vectors optimized with the proposed method provide higher recognition rates with the same size vectors and comparable results even with lower dimensional vectors. We also compare the proposed algorithm to more recent texture recognition methods based on convolutional neural networks and show that it can still provide comparable results even though the resulting feature vectors are smaller by orders of magnitude.
... Alexandre [10] combined LBP with intensity and shape features using a multiscale fusion approach. In [11], LBP is combined with contrast information for gender classification where local contrast histograms are used as [12] used statistical features such as mean, variance, skew and kurtosis together with LBP. Different enhancements inspired by LBP have been proposed to solve the problem of gender recognition [13], [14], [15], [16]. ...
... This loss of information reduces the discrimination ability of the LBP operator. To handle this issue, Ylioinas et al. [27] proposed to combine contrast information with LBP and showed improved gender classification performance. The other two relevant works namely extended local binary pattern (ELBP) [28] and completed local binary patterns (CLBP) [29] followed a different approach to utilize GLD magnitude information. ...
... However, the information about the magnitude of the difference is completely lost in the process which limits the discrimination capability of LBP. To overcome this limitation, Ylioinas et al. [27] proposed to combine LBP with contrast information to improve gender classification performance. The contrast information was computed using the local variance measure (VAR) as follows: ...
... Before presenting experimental results, we would like to highlight the differences between the proposed work and the work in [27]. To make comparison easier, the differences are tabulated in Table 1. ...
This paper presents multi-quantized local binary patterns for facial gender classification. For encoding the gray level difference (GLD) between a reference pixel and its neighbors, local binary pattern employs a binary quantization which retains the sign of GLD but discards the magnitude information. To improve the discrimination capability, the proposed method utilizes both the sign and magnitude components by performing multi-level vector quantization of GLD. Each quantized level is then separately encoded to generate multiple local binary patterns. The proposed method is evaluated on four publicly available datasets (FERET, PAL, CASIA and FEI) through extensive experiments. Comparison of performance with various existing methods clearly demonstrated that the proposed method has advantages such as higher discrimination power, improved noise robustness and better generalization capability.
... First, a larger than 3x3 pixels square neighbourhood can be used (usually in the form of a circle with two parameters P -number of neighbours and R -radius). Second, the so-called uniform LBP [7] can be used. A binary number is uniform if it has no more than two transitions from 0 to 1 or 1 to 0. The uniform patterns are the most informative ones, therefore they were used in the described application. ...
In this paper the possibility of estimating age, gender and race of people based on visual information - facial images - is analysed. The face detection is performed using the Viola-Jones approach, for face alignment the affine transform is used and as feature vector histograms computed for non-overlapping windows for the LBP (Local Binary Pattern) descriptor image are used. The following accuracies were obtained: over 90 % for gender estimation, over 86% for race estimation and 70% for age estimation
... The work of Tan and Triggs [38] combines Gabor wavelets and LBP for the problem of face recognition. Ylioinas et al. [46] combine contrast information together with local binary patterns for improved gender classification. A combination of HOG, LBP and Gabor features is used by Li et al. [23] for gender classification. ...
... This extends the applicability of our platform. Also for these purposes, methods based on LBP variants were adopted in our system [22,23]. ...
A service robot should be able to detect and identify the user in order to personalize its services and guarantee security, it should recognize the user’s emotions to allow affective interaction, and it should be able to communicate easily with the user and understand given commands by recognizing speech and gestures. Our research is motivated, for example, by the emerging needs of elderly care, health care, safety, and logistics. We have developed a distributed system for affective human–robot interaction which combines all the basic sensory elements and can collaborate with a smart environment and obtain further knowledge from the Internet. The realized HRI robot, Minotaurus, runs in real time and is capable of interacting with human beings. Minotaurus forms a rather generic platform for experimenting with human–robot collaboration in different applications and environments.
... The original LBP has shown powerful and effective results against variations in rotation and illumination. Recently, it has been widely used in various studies and applications (Houam et al., 2012;Tlioinas et al., 2011;Kim et al., 2012). The LBP operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. ...
... The LBP operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. It is a powerful means of texture description; among its properties in real-world applications one can mention its discriminative power, computational simplicity and tolerance against monotonic gray-scale changes (Ahonen et al., 2006;Pietikainen et al., 2010;Tlioinas et al., 2011). ...
In this study of a biometric system, Improved One Dimensional Local Binary Patterns
(I1DLBP) are developed and tested for use in face analysis, description and recognition. The extraction
of facial features is based on the principal that the human visual system combines local and global
features to differentiate between people. The proposed method starts by decomposing the facial image
into several blocks with different resolutions. Each block is then projected in one dimensional space, and
the developed descriptor is applied on each projected block. Finally, Principal Component Analysis
(PCA) is used to reduce the dimensionalities of the concatenated vectors from each block and to keep
only the relevant information. The K-nearest neighbors (KNN) algorithm is used as a classifier.
Experiments were carried out under varying conditions of occlusion, rotation, and facial expressions,
using the ORL and AR databases. Results show that the developed feature extraction approach can
effectively describe the micro characteristics of the human face and that it outperforms well-known and
classical feature extraction descriptors.
... The original LBP has shown powerful and effective results against variations in rotation and illumination. Recently, it has been widely used in various studies and applications (Houam et al., 2012;Tlioinas et al., 2011;Kim et al., 2012). The LBP operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. ...
... The LBP operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. It is a powerful means of texture description; among its properties in real-world applications one can mention its discriminative power, computational simplicity and tolerance against monotonic gray-scale changes (Ahonen et al., 2006;Pietikainen et al., 2010;Tlioinas et al., 2011). ...
In this study of biometric system, Improved One Dimensional Local Binary Patterns (I1DLBP) are developed and tested for use in face analysis, description and recognition. The extraction of face’s features is inspired from the principal that the human visual system combines between local and global features to differentiate between peoples. It starts by decomposing the facial image into several blocks with different resolutions. Then, each block is projected in one dimensional space. Next, the developed descriptor is applied on each projected block. Finally, Principal Component Analysis (PCA) is needed to reduce the dimensionalities of the concatenated vectors from each block and to keep only the pertinent information. K-nearest neighbor (KNN) is used as a classifier. Experiments were carried out under varying conditions in occlusion, rotation and, facial expressions using ORL and AR databases. Results show that the developed feature extraction approach can effectively describe the micro characteristics of the human face and the superiority in comparison to well-known and classical feature extraction descriptors.
... Since then, significant progress has been reported in the literature. Fundamentally, proposed techniques differ in (i) face images-2D or 3D (ii) choice of facial representation, ranging from simple raw 2D/3D pixels to complex features and (iii) design of classifiers, such as Neural Networks, SVM and boosting methods [2]. ...
Age reflects the continuous accumulation of durable effects from the past since birth. Human faces deform with time non-inversely and thus contains their aging information. In addition to its richness with anatomy information, 3D shape of faces could have the advantage of less dependent on pose and independent of illumination, while it hasn't been noticed in literature. Thus, in this work we investigate the age estimation problem from 3D shape of the face. With several descriptions grounding on Riemannian shape analysis of facial curves, we first extracted features from ideas of face Averageness, face Symmetry, its shape variations with Spatial and Gradient descriptors. Then, using the Random Forest-based Regression, experiments are carried out following the Leaving-One-Person-Out (LOPO) protocol on the FRGCv2 dataset. The proposed approach performs with a Mean Absolute Error (MAE) of 3:29 years using a gender-general test protocol. Finally, with the gender-specific experiments, which first separate the 3D scans into Female and Male subsets, then train and test on each gender specific subset in LOPO fashion, we improves the MAE to 3:15 years, which confirms the idea that the aging effect differs with gender. Copyright © 2014 SCITEPRESS - Science and Technology Publications. All rights reserved.
