An Efficient Face Recognition System Based On the Hybridization of Pose Invariant and Illumination Process

Documento informativo
IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 4, No 1, July 2012 ISSN (Online): 1694-0814 228 S. Muruganantham† and Dr. T. Jebarajan††, Assistant Professor, S.T.Hindu College, Nagercoil. †† Principal, Kings College of Engineering, Chennai. † Abstract: In the previous decade, one of the most effectual applications of image analysis and indulgent that attracted significant consideration is the human face recognition. One of the diverse techniques used for identifying an individual is the Face recognition. Normally the image variations for the reason that of the change in face identity are less than the variations between the images of the same face under different illumination and viewing angle. Among several factors that manipulate face recognition, illumination and pose are the two major challenges. Pose and illumination variations harshly affect the performance of face recognition. Considerably less effort has been taken to deal with the problem of mutual variations of pose and illumination in face recognition, while several algorithms have been proposed for face recognition from fixed points. In this paper we intend a face recognition method that is forceful to pose and illumination variations. We first put forward a simple pose estimation method based on 2D images, which uses a proper classification rule and image representation to classify a pose of a face image. After that, the image can be assigned to a pose class by a classification rule in a low-dimensional subspace constructed by a feature extraction method. We offer a shadow compensation method that compensates for illumination variation in a face image so that the image can be predictable by a face recognition system designed for images under normal illumination condition. Starting the accomplishment result, it is obvious that our projected technique based on the hybridization system recognizes the face images effectively. Keywords: Face recognition, Pose, illumination, Edge detection, Shadow compensation 1. Introduction The increasing significance of safety and group surrounded by various platforms presently offered credentials and validation methods crooked out to key technology in different areas [1]. In recent times, the exploit of biometrics has improved considerably in personal security and/or right to use control applications [3]. Biometrics technology is likely to replace conventional validation methods that are easily stolen, elapsed and duplicated. Biometrics applied in fingerprints and iris scan deliver extremely steadfast performance, however human face reside as an striking biometric due to its payback over some of the other biometrics. Face recognition is non-intrusive and it needs no support from the test subjects, whereas supplementary biometrics necessitates a subject’s cooperation. In favor of case, inside iris or fingerprint recognition, one be supposed to have a fleeting look over an eye scanner or place their finger on a fingerprint reader [4]. In the face of few biometric methods to facilitate the intrinsic worth of both high accuracy and low meddlingness, Face Recognition Technology has varied prospective over applications in the fields of information security, law enforcement and inspection, smart cards, access control and more [5], [2], [6]. Face recognition seems to encompass aspire challenges, while faces of different persons throw in to global shape individuality, whereas face images of a single person is subject to considerable variations, which might overcome the measured inter-person differences Such variation is owing to the facts to incorporate facial expressions, illumination conditions, pose, presence or absence of eyeglasses and facial hair, occlusion, and aging [7, 8]. The essential consequence of face recognition as a biometric is its throughput, expediency and noninvasiveness. Most of the face recognition investigation done till date uses 2D intensity images as the data format for processing. Success has been achieved at diverse levels over 2D face recognition research [9]. The two major categories of face recognition scenario i.e. Face verification is a conversation match that evaluates a query of face image in counter with a gallery face image (The gallery face images have been stored in the database.) whose exceptionality being uncovered. Face identification is one-to-many matching processes that reflect on a query face image against all the gallery images in a face database to understand the individuality of the query face. The exposure of the query image is delivered by spotting the image in the database that has the maximum resemblance with the query image [10]. Face recognition seems to be the major issue in abundant notorious regions of machine vision for about a decennium. Topical systems have progressed to be reasonably precise in recognition below controlled situation. Conversely extrinsic imaging parameters such as pose, illumination, and facial expression still cause much complication in correct recognition [11]. In general, human face is similar in shape, have two eyes, a mouth and a nose. Copyright (c) 2012 International Journal of Computer Science Issues. All Rights Reserved. IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 4, No 1, July 2012 ISSN (Online): 1694-0814 Every component makes different distinctive shadows and specularities depending on the direction of the lighting in a fixed pose. By means of such distinctive, the lighting direction can be estimated and illumination variation can be remunerated. Likewise, in a convenient application environment, the illumination variation is constantly tied with other problem such as pose variation, which increases the convolution of the automatic face recognition problem. Changes in screening conditions or rotations of faces carry difficulties to a face recognition system. There are two types of face rotations to be considered; first one is the rotation of human face images in the imageplane and the second one is the rotation of faces out of the image plane (in-depth rotation). In the former case, face images can be simply standardized by detecting several pointers in the face after that applying some essential transformations, while in the second case such standardization is inaccessible even as some parts of faces may be occluded. At a halt for the same person, huge variation of look is caused by pose variation. The variation is frequently more important than that warped by different persons under the same pose [21]. Hence the humiliation of conservative appearance-based methods for instance Eigen face etc., humiliates severely when non frontal searches match against the registered frontal faces [20]. Amidst the several methods proposed to address the pose problem the extensively used method are view-based methods [19]. Numeral illuminations constant face recognition looms have been projected in the past years. Presented approaches deal with the illumination variation problem drop into two main categories. The first category is “passive” approaches, as they effort to conquer this problem by studying the visible spectrum images in which face appearance has been changed by illumination variations. The other category contains “active” approaches, in which the illumination variation problem is conquer by employing active imaging techniques to obtain face images captured in steady illumination condition, or images of illumination invariant modalities. The changes stimulated by illumination are frequently larger than the differences between individuals, causing systems based directly on comparing images to misclassify input images. There are four main categories of looms for handling variable illumination: (1) extraction of illumination constant aspects (2) renovation of images with uneven illuminations to a canonical representation (3) modeling the illumination variations (4) exploitation of some 3D face models whose facial shapes and albedos are obtained in advance [7]. One of the most significant troubles in face recognition is the variable illumination. The key fact is that illumination, along with pose variation, is the most important factor that changes the observation (appearance) 229 of faces. Lighting conditions vary mostly between indoor and outdoor environments, but also within indoor environments. Hence, owing to the 3D shape of human faces, a direct lighting source can produce strong shadows that highlight or shrink certain facial aspects. In excess of some previous decades, many face recognition methods have been developed. Frequently used feature extraction methods are chief component analysis (PCA) [12], [13] and linear discriminant analysis (LDA) [14], [15], [16], [17]. Another linear technique identified as Locality Preserving Projections (LPP) [18], [16], which finds an embedding that conserve local information, and expands a face subspace that best detects the essential face diverse structure. Out of the various factors that involve face recognition Illumination and pose are the two main confronts. Pose and lighting variations significantly deteriorate the utility of recognition. So to undertake this problem, the face is recognized under various lighting conditions by our projected system. The respite of the proposal is ordered as follows: section 2 briefs a few recent related works. Section 3 briefs the face recognition process based on a hybrid approach. Tentative consequences and study of the projected methodology are discussed in Section 4. In conclusion, remarks are provided in Section 5. 2. Literature review Some of the face recognition schemes, which make use of pattern toning and model based techniques for better performance, have been presented in the literature. Newly, incorporating pose invariant techniques into face recognition schemes to improve its performance and efficiency has expected a great deal of consideration among researchers in face recognition community. A pithy appraisal of a few current researches is presented here. Wide-ranging hard work have been put into their search toward pose-invariant face recognition in recent years and many prominent approaches have been proposed by Xiao Zheng Zhang and Yong Sheng Gao [22]. Their paper provides a decisive analysis of researches on image-based face recognition across pose. The presented techniques were broadly reviewed and discussed. These techniques were classified into different categories according to the methodologies in handling pose variations. Their approaches, advantages/disadvantages and performances were convoluted. By simplifying diverse strategy in handling pose variations and appraising their performances, several capable directions for future research have been suggested. Arashloo and Kittler [30] have tackled the problem of face recognition under random pose. A hierarchical MRFbased image identical technique for finding pixel-wise communications between facial images viewed from different angles was planned and used to closely register a Copyright (c) 2012 International Journal of Computer Science Issues. All Rights Reserved. IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 4, No 1, July 2012 ISSN (Online): 1694-0814 pair of facial images. The goodness-of-match between two faces was subsequently measured in terms of the standardized energy of the match which has a combination of both structural differences between faces as well as their texture uniqueness. The method needs no training on nonfrontal images and circumvents the need for geometrical normalization of facial images. It was also forceful to moderate scale changes between images. Dahm and Yongsheng Gao [31] have explained that several Face Recognition techniques have focal point on 2D- 2D comparison or 3D-3D comparison; but only some techniques survey the idea of cross-dimensional assessment. Their paper offered a new face recognition approach that outfit cross-dimensional assessment to solve the problem of pose invariance. Their approach implements a Gabor representation during comparison to allow for variations in texture, illumination, expression and pose. Kernel scaling was used to shrink comparison time during the branching search, which establishes the facial pose of input images. In their paper, they present a novel face recognition approach that utilizes 3D data to conquer changes in facial pose, while remaining noninterfering. To realize this, we use 3D textured head models in the gallery, as gallery data is generally taken from supportive subjects (e.g. identification photos, mug shots). For the query, we use 2D images, which can be taken from submissive cameras such as ceiling mounted surveillance cameras. Together this gives us a crossdimensional approach, combined with a non-interfering nature. An efficient novel H-eigenface (Hybrid-eigenface) system for pose invariant face recognition varying from fore to summary view has been offered by Abhishek Sharma et al [32]. H-eigenfaces were completely new source for face image representation in different poses and are used for effective fore view separation. The projected method was based on the fact that face samples of same person under different poses are identical in terms of the permutation pattern of facial features. H-eigenfaces develop that fact and thus two H-eigenfaces under different poses capture same features of the face. Thus providing a solid view-based subspace, which can be more used to generate virtual frontal view from inputted nonfrontal face image using least square projection technique? The use of their proposed tactic on FERET and ORL face database shows an exciting improvement in recognition accuracy and a discrete reduction in online computation when compared to global linear regression method. Real-time pose invariant face recognition algorithms from an arcade of frontal images have been projected by Choi Hyun Chul and Oh Se-Young [33]. Initially, they customized the second order minimization method for active appearance model (AAM). That tolerates the AAM to have the ability of correct union with small loss of frame rate. Next, they encompass future pose transforming 230 matrix which can eradicate warping object of the warped face image from AAM fitting. That makes it possible to prepare a neural network as the face recognizer with one frontal face image of each person in the gallery set. Third, they recommend an effortless method for pose recognition by means of neural networks to select suitable pose transforming matrix. Khaleghian and Rohban [34] have projected an ensemble-based loom to enhance concert of Tied Factor Analysis (TFA) to conquer some of the disputes in face recognition across large pose variations. They used Adaboost to boost TFA which has shown to possess stateof-the-art face recognition performance under large pose variations. They have employed boosting as a discriminative training in the TFA as a generative model. In their model, TFA was used as a base classifier for the boosting algorithm and a biased probability model for TFA was proposed to adjust the importance of each training data. Likewise, a customized weighting and a variety measure are used to produce more diverse classifiers in the boosting process. Experimental results on the FERET data set confirmed the improved performance of the Boosted Tied Factor Analysis (BTFA) in comparison with TFA for lower dimensions when a holistic approach was used. 3. An Efficient Face Recognition System Let D be a database containing N number of images, {I1 , I 2 ,., I N }and let I be a database image of size R × S . D , After inputting an image from the database D , the user must specify the type of the image i.e., whether it is pose invariant, illumination invariant or both pose and illumination variant. Depending on the image group precise by the user, any one of the following three processes is executed by the system. 3.1 Face Detection In the recent years face detection have attained to a great extent. Various approaches were offered in excess of previous years in the field of face and eye detection. The position of the art techniques are emergence based methods, which include a lot of ways for object recognition such as neural networks (NN), support vector machines (SVM) etc. The beginning of face detection using neural network, by Rowley et al, in 1996, was an important work by that time [20]. Consequently, there are many other approaches. Interested reader may read the comprehensive survey paper by Yang, et al [21]. Viola and Jones [22] uses boosted cascade of simple Haar-like features introduced by Papageorgiou [24] and enhanced by Viola [22] and Lienhart [23] for object detection. This is one of the most discernible algorithms in face detection. Copyright (c) 2012 International Journal of Computer Science Issues. All Rights Reserved. IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 4, No 1, July 2012 ISSN (Online): 1694-0814 The Viola-Jones [22] face detection system makes use of a multi-scale multi-stage classifier that functions on image intensity information. This manner of face detection is operated in the existing system that doesn’t require any previous calibration process. In general this loom scrolls a window across the image and consigns a binary classifier that discriminates between a face and the background. This classifier is regimented with a boosting machine learning meta-algorithm. It is endorsed as the fastest and most accurate pattern recognition method for faces in monocular grey-level images. They urbanized a real-time face detector comprising of a cascade of classifiers trained by AdaBoost. Every classifier workout with an integral image filter, which hark back of Haar Basis functions and can be processed very fast at any location and scale. This is important to speed up the detector. At equilibrium value of MeCpG steps (,+14 deg.) [31,44]. In comparison, methylation has a significantly lower stability cost when happening at major groove positions, such as 211 and 21 base pair from dyad (mutations 9 and 12), where the roll of the nucleosome bound conformation (+10 deg.) is more compatible with the equilibrium geometry of MeCpG steps. The nucleosome destabilizing effect of cytosine methylation increases with the number of methylated cytosines, following the same position dependence as the single methylations. The multiple-methylation case reveals that each major groove meth- PLOS Computational Biology | 3 November 2013 | Volume 9 | Issue 11 | e1003354 DNA Methylation and Nucleosome Positioning ylation destabilizes the nucleosome by around 1 kJ/mol (close to the average estimate of 2 kJ/mol obtained for from individual methylation studies), while each minor groove methylation destabilizes it by up to 5 kJ/mol (average free energy as single mutation is around 6 kJ/mol). This energetic position-dependence is the reverse of what was observed in a recent FRET/SAXS study [30]. The differences can be attributed to the use of different ionic conditions and different sequences: a modified Widom-601 sequence of 157 bp, which already contains multiple CpG steps in mixed orientations, and which could assume different positioning due to the introduction of new CpG steps and by effect of the methylation. The analysis of our trajectories reveals a larger root mean square deviation (RMSD) and fluctuation (RMSF; see Figures S2– S3 in Text S1) for the methylated nucleosomes, but failed to detect any systematic change in DNA geometry or in intermolecular DNA-histone energy related to methylation (Fig. S1B, S1C, S4–S6 in Text S1). The hydrophobic effect should favor orientation of the methyl group out from the solvent but this effect alone is not likely to justify the positional dependent stability changes in Figure 2, as the differential solvation of the methyl groups in the bound and unbound states is only in the order of a fraction of a water molecule (Figure S5 in Text S1). We find however, a reasonable correlation between methylation-induced changes in hydrogen bond and stacking interactions of the bases and the change in nucleosome stability (see Figure S6 in Text S1). This finding suggests that methylation-induced nucleosome destabilization is related to the poorer ability of methylated DNA to fit into the required conformation for DNA in a nucleosome. Changes in the elastic deformation energy between methylated and un-methylated DNA correlate with nucleosomal differential binding free energies To further analyze the idea that methylation-induced nucleosome destabilization is connected to a worse fit of methylated DNA into the required nucleosome-bound conformation, we computed the elastic energy of the nucleosomal DNA using a harmonic deformation method [36,37,44]. This method provides a rough estimate of the energy required to deform a DNA fiber to adopt the super helical conformation in the nucleosome (full details in Suppl. Information Text S1). As shown in Figure 2, there is an evident correlation between the increase that methylation produces in the elastic deformation energy (DDE def.) and the free energy variation (DDG bind.) computed from MD/TI calculations. Clearly, methylation increases the stiffness of the CpG step [31], raising the energy cost required to wrap DNA around the histone octamers. This extra energy cost will be smaller in regions of high positive roll (naked DNA MeCpG steps have a higher roll than CpG steps [31]) than in regions of high negative roll. Thus, simple elastic considerations explain why methylation is better tolerated when the DNA faces the histones through the major groove (where positive roll is required) that when it faces histones through the minor groove (where negative roll is required). Nucleosome methylation can give rise to nucleosome repositioning We have established that methylation affects the wrapping of DNA in nucleosomes, but how does this translate into chromatin structure? As noted above, accumulation of minor groove methylations strongly destabilizes the nucleosome, and could trigger nucleosome unfolding, or notable changes in positioning or phasing of DNA around the histone core. While accumulation of methylations might be well tolerated if placed in favorable positions, accumulation in unfavorable positions would destabilize the nucleosome, which might trigger changes in chromatin structure. Chromatin could in fact react in two different ways in response to significant levels of methylation in unfavorable positions: i) the DNA could either detach from the histone core, leading to nucleosome eviction or nucleosome repositioning, or ii) the DNA could rotate around the histone core, changing its phase to place MeCpG steps in favorable positions. Both effects are anticipated to alter DNA accessibility and impact gene expression regulation. The sub-microsecond time scale of our MD trajectories of methylated DNAs bound to nucleosomes is not large enough to capture these effects, but clear trends are visible in cases of multiple mutations occurring in unfavorable positions, where unmethylated and methylated DNA sequences are out of phase by around 28 degrees (Figure S7 in Text S1). Due to this repositioning, large or small, DNA could move and the nucleosome structure could assume a more compact and distorted conformation, as detected by Lee and Lee [29], or a slightly open conformation as found in Jimenez-Useche et al. [30]. Using the harmonic deformation method, we additionally predicted the change in stability induced by cytosine methylation for millions of different nucleosomal DNA sequences. Consistently with our calculations, we used two extreme scenarios to prepare our DNA sequences (see Fig. 3): i) all positions where the minor grooves contact the histone core are occupied by CpG steps, and ii) all positions where the major grooves contact the histone core are occupied by CpG steps. We then computed the elastic energy required to wrap the DNA around the histone proteins in unmethylated and methylated states, and, as expected, observed that methylation disfavors DNA wrapping (Figure 3A). We have rescaled the elastic energy differences with a factor of 0.23 to match the DDG prediction in figure 2B. In agreement with the rest of our results, our analysis confirms that the effect of methylation is position-dependent. In fact, the overall difference between the two extreme methylation scenarios (all-in-minor vs all-in-major) is larger than 60 kJ/mol, the average difference being around 15 kJ/ mol. We have also computed the elastic energy differences for a million sequences with CpG/MeCpG steps positioned at all possible intermediate locations with respect to the position (figure 3B). The large differences between the extreme cases can induce rotations of DNA around the histone core, shifting its phase to allow the placement of the methylated CpG steps facing the histones through the major groove. It is illustrative to compare the magnitude of CpG methylation penalty with sequence dependent differences. Since there are roughly 1.5e88 possible 147 base pairs long sequence combinations (i.e., (4n+4(n/2))/2, n = 147), it is unfeasible to calculate all the possible sequence effects. However, using our elastic model we can provide a range of values based on a reasonably large number of samples. If we consider all possible nucleosomal sequences in the yeast genome (around 12 Mbp), the energy difference between the best and the worst sequence that could form a nucleosome is 0.7 kj/mol per base (a minimum of 1 kJ/mol and maximum of around 1.7 kJ/mol per base, the first best and the last worst sequences are displayed in Table S3 in Text S1). We repeated the same calculation for one million random sequences and we obtained equivalent results. Placing one CpG step every helical turn gives an average energetic difference between minor groove and major groove methylation of 15 kJ/ mol, which translates into ,0.5 kJ/mol per methyl group, 2 kJ/ mol per base for the largest effects. Considering that not all nucleosome base pair steps are likely to be CpG steps, we can conclude that the balance between the destabilization due to CpG methylation and sequence repositioning will depend on the PLOS Computational Biology | 4 November 2013 | Volume 9 | Issue 11 | e1003354 DNA Methylation and Nucleosome Positioning Figure 3. Methylated and non-methylated DNA elastic deformation energies. (A) Distribution of deformation energies for 147 bplong random DNA sequences with CpG steps positioned every 10 base steps (one helical turn) in minor (red and dark red) and major (light and dark blue) grooves respectively. The energy values were rescaled by the slope of a best-fit straight line of figure 2, which is 0.23, to source of circulating FGF-21. The lack of association between circulating and muscle-expressed FGF-21 also suggests that muscle FGF-21 primarily works in a local manner regulating glucose metabolism in the muscle and/or signals to the adipose tissue in close contact to the muscle. Our study has some limitations. The number of subjects is small and some correlations could have been significant with greater statistical power. Another aspect is that protein levels of FGF-21 were not determined in the muscles extracts, consequently we cannot be sure the increase in FGF-21 mRNA is followed by increased protein expression. In conclusion, we show that FGF-21 mRNA is increased in skeletal muscle in HIV patients and that FGF-21 mRNA in muscle correlates to whole-body (primarily reflecting muscle) insulin resistance. These findings add to the evidence that FGF-21 is a myokine and that muscle FGF-21 might primarily work in an autocrine manner. Acknowledgments We thank the subjects for their participation in this study. Ruth Rousing, Hanne Willumsen, Carsten Nielsen and Flemming Jessen are thanked for excellent technical help. The Danish HIV-Cohort is thanked for providing us HIV-related data. PLOS ONE | 6 March 2013 | Volume 8 | Issue 3 | e55632 Muscle FGF-21,Insulin Resistance and Lipodystrophy Author Contributions Conceived and designed the experiments: BL BKP JG. Performed the experiments: BL TH TG CF PH. Analyzed the data: BL CF PH. Contributed reagents/materials/analysis tools: BL. Wrote the paper: BL. References 1. Kharitonenkov A, Shiyanova TL, Koester A, Ford AM, Micanovic R, et al. (2005) FGF-21 as a novel metabolic regulator. J Clin Invest 115: 1627–1635. 2. Coskun T, Bina HA, Schneider MA, Dunbar JD, Hu CC, et al. (2008) Fibroblast growth factor 21 corrects obesity in mice. Endocrinology 149: 6018– 6027. 3. Xu J, Lloyd DJ, Hale C, Stanislaus S, Chen M, et al. (2009) Fibroblast growth factor 21 reverses hepatic steatosis, increases energy expenditure, and improves insulin sensitivity in diet-induced obese mice. Diabetes 58: 250–259. 4. Inagaki T, Dutchak P, Zhao G, Ding X, Gautron L, et al. (2007) Endocrine regulation of the fasting response by PPARalpha-mediated induction of fibroblast growth factor 21. Cell Metab 5: 415–425. 5. Potthoff MJ, Inagaki T, Satapati S, Ding X, He T, et al. (2009) FGF21 induces PGC-1alpha and regulates carbohydrate and fatty acid metabolism during the adaptive starvation response. Proc Natl Acad Sci U S A 106: 10853–10858. 6. Badman MK, Pissios P, Kennedy AR, Koukos G, Flier JS, et al. (2007) Hepatic fibroblast growth factor 21 is regulated by PPARalpha and is a key mediator of hepatic lipid metabolism in ketotic states. Cell Metab 5: 426–437. 7. Zhang X, Yeung DC, Karpisek M, Stejskal D, Zhou ZG, et al. (2008) Serum FGF21 levels are increased in obesity and are independently associated with the metabolic syndrome in humans. Diabetes 57: 1246–1253. 8. Chen WW, Li L, Yang GY, Li K, Qi XY, et al. (2008) Circulating FGF-21 levels in normal subjects and in newly diagnose patients with Type 2 diabetes mellitus. Exp Clin Endocrinol Diabetes 116: 65–68. 9. Chavez AO, Molina-Carrion M, Abdul-Ghani MA, Folli F, DeFronzo RA, et al. (2009) Circulating fibroblast growth factor-21 is elevated in impaired glucose tolerance and type 2 diabetes and correlates with muscle and hepatic insulin resistance. Diabetes Care 32: 1542–1546. 10. Hojman P, Pedersen M, Nielsen AR, Krogh-Madsen R, Yfanti C, et al. (2009) Fibroblast growth factor-21 is induced in human skeletal muscles by hyperinsulinemia. Diabetes 58: 2797–2801. 11. Izumiya Y, Bina HA, Ouchi N, Akasaki Y, Kharitonenkov A, et al. (2008) FGF21 is an Akt-regulated myokine. FEBS Lett 582: 3805–3810. 12. Vienberg SG, Brons C, Nilsson E, Astrup A, Vaag A, et al. (2012) Impact of short-term high-fat feeding and insulin-stimulated FGF21 levels in subjects with low birth weight and controls. Eur J Endocrinol 167: 49–57. 13. Carr A, Samaras K, Burton S, Law M, Freund J, et al. (1998) A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS 12: F51–F58. 14. Haugaard SB, Andersen O, Dela F, Holst JJ, Storgaard H et al. (2005) Defective glucose and lipid metabolism in human immunodeficiency virus-infected patients with lipodystrophy involve liver, muscle tissue and pancreatic betacells. Eur J Endocrinol 152: 103–112. 15. Reeds DN, Yarasheski KE, Fontana L, Cade WT, Laciny E, et al. (2006) Alterations in liver, muscle, and adipose tissue insulin sensitivity in men with HIV infection and dyslipidemia. Am J Physiol Endocrinol Metab 290: E47–E53. 16. Meininger G, Hadigan C, Laposata M, Brown J, Rabe J, et al. (2002) Elevated concentrations of free fatty acids are associated with increased insulin response to standard glucose challenge in human immunodeficiency virus-infected subjects with fat redistribution. Metabolism 51: 260–266. 17. Carr A, Emery S, Law M, Puls R, Lundgren JD, et al. (2003) An objective case definition of lipodystrophy in HIV-infected adults: a case-control study. Lancet 361: 726–735. 18. Lindegaard B, Hansen T, Hvid T, van HG, Plomgaard P, et al. (2008) The effect of strength and endurance training on insulin sensitivity and fat distribution in human immunodeficiency virus-infected patients with lipodystrophy. J Clin Endocrinol Metab 93: 3860–3869. 19. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, et al. (1985) Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28: 412– 419. 20. Lindegaard B, Frosig C, Petersen AM, Plomgaard P, Ditlevsen S, et al. (2007) Inhibition of lipolysis stimulates peripheral glucose uptake but has no effect on endogenous glucose production in HIV lipodystrophy. Diabetes 56: 2070–2077. 21. DeFronzo RA, Tobin JD, Andres R (1979) Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 237: E214–E223. 22. Plomgaard P, Bouzakri K, Krogh-Madsen R, Mittendorfer B, Zierath JR, et al. (2005) Tumor necrosis factor-alpha induces skeletal muscle insulin resistance in healthy human subjects via inhibition of Akt substrate 160 phosphorylation. Diabetes 54: 2939–2945. 23. Thomas JA, Schlender KK, Larner J (1968) A rapid filter paper assay for UDPglucose-glycogen glucosyltransferase, including an improved biosynthesis of UDP-14C-glucose. Anal Biochem 25: 486–499. 24. Haugaard SB, Andersen O, Madsbad S, Frosig C, Iversen J, et al. (2005) Skeletal Muscle Insulin Signaling Defects Downstream of Phosphatidylinositol 3-Kinase at the Level of Akt Are Associated With Impaired Nonoxidative Glucose Disposal in HIV Lipodystrophy. Diabetes 54: 3474–3483. 25. Boden G, Jadali F, White J, Liang Y, Mozzoli M, et al. (1991) Effects of fat on insulin-stimulated carbohydrate metabolism in normal men. J Clin Invest 88: 960–966. 26. Mashili FL, Austin RL, Deshmukh AS, Fritz T, Caidahl K, et al. (2011) Direct effects of FGF21 on glucose uptake in human skeletal muscle: implications for type 2 diabetes and obesity. Diabetes Metab Res Rev 27: 286–297. 27. Torriani M, Thomas BJ, Barlow RB, Librizzi J, Dolan S, et al. (2006) Increased intramyocellular lipid accumulation in HIV-infected women with fat redistribution. J Appl Physiol 100: 609–614. 28. Lee MS, Choi SE, Ha ES, An SY, Kim TH, et al. (2012) Fibroblast growth factor-21 protects human skeletal muscle myotubes from palmitate-induced insulin resistance by inhibiting stress kinase and NF-kappaB. Metabolism . 29. Tyynismaa H, Carroll CJ, Raimundo N, Ahola-Erkkila S, Wenz T, et al. (2010) Mitochondrial myopathy induces a starvation-like response. Hum Mol Genet 19: 3948–3958. 30. Maagaard A, Holberg-Petersen M, Kollberg G, Oldfors A, Sandvik L, et al. (2006) Mitochondrial (mt)DNA changes in tissue may not be reflected by depletion of mtDNA in peripheral blood mononuclear cells in HIV-infected patients. Antivir Ther 11: 601–608. 31. Payne BA, Wilson IJ, Hateley CA, Horvath R, Santibanez-Koref M, et al. (2011) Mitochondrial aging is accelerated by anti-retroviral therapy through the clonal expansion of mtDNA mutations. Nat Genet 43: 806–810. 32. Kliewer SA, Mangelsdorf DJ (2010) Fibroblast growth factor 21: from pharmacology to physiology. Am J Clin Nutr 91: 254S–257S. 33. Gallego-Escuredo JM, Domingo P, Gutierrez MD, Mateo MG, Cabeza MC, et al. (2012) Reduced Levels of Serum FGF19 and Impaired Expression of Receptors for Endocrine FGFs in Adipose Tissue From HIV-Infected Patients. J Acquir Immune Defic Syndr 61: 527–534. 34. Domingo P, Gallego-Escuredo JM, Domingo JC, Gutierrez MM, Mateo MG, et al. (2010) Serum FGF21 levels are elevated in association with lipodystrophy, insulin resistance and biomarkers of liver injury in HIV-1-infected patients. AIDS 24: 2629–2637. PLOS ONE | 7 March 2013 | Volume 8 | Issue 3 | e55632
An Efficient Face Recognition System Based On the Hybridization of Pose Invariant and Illumination Process
Documento similar
Face Recognition



Edge Detection

Shadow Compensation


An Efficient Face Recognition System Based On the Hybridization of Pose Invariant and Illumination Process