Cette thèse s’inscrit dans le domaine de l’authentification de visage, en particulier dans le cadre du projet national « TechnoVision ». Bien que les êtres humains puissent détecter/reconnaître des visages dans une scène sans beaucoup de peine, construire un système qui accomplit de telles tâches représente un sérieux défie. Ce défi est d’autant plus grand lorsque les conditions d’acquisition des images sont très variables. Il existe deux types de variations associées aux images de visages : inter et intra sujet. La variation inter-sujet est limitée à cause du fait que la ressemblance physique entre les individus est assez rare. Par contre la variation intra-sujet est plus courante à cause, par exemple, des variations de poses, des changements dans les conditions d’éclairage, etc. Dans cette thèse, nous avons d’abord développé une approche de localisation de visage et de ses caractéristiques pour des images ne contenant qu’un seul visage sur un fond relativement uniforme avec des variations de lumière. Pour cela nous avons proposé une approche robuste de segmentation couleur dans l’espace TLS qui repose sur l’algorithme de ligne de partage des eaux modifiée. Pour l’extraction des régions caractéristiques faciales (yeux et bouche), nous avons combiné la méthode de classification kmeans avec une approche géométrique et nous l’avons appliqué sur la région du visage segmentée. Nous avons aussi proposé une approche multimodale 2D/3D qui repose sur la fusion pondérée des scores de l’approche « EigenFace » modulaire avec la signature anthropométrique 3D de visage. Nous avons évalué nos approches 3D et 2D/3D de reconnaissance du visage sur une sous-base de IV2 qui contient des images stéréoscopiques de visage. Les résultats obtenus sont très intéressants en comparaison avec les techniques classiques de reconnaissance 2D de visage. Enfin, nous avons discuté les perspectives d’amélioration des approches proposées. / This thesis deals with the face authentification problem, in particular within a national project framework, namely "TechnoVision". Although the human beings can detect/recognise faces in a scene without much of sorrow, build a system which achieves such tasks is very challenging. This challenge is all the more large when the conditions of images acquisition are variable. There are two kinds of variations associated to the face images: inter and intra subject. The inter-subject variation is limited because owing to the fact that the physical resemblance between the individuals is rather rare. On the other hand, the intra-subject variation is more current because of pose changing, lighting conditions, etc. In this thesis, we developed, first, an approach for face and facial features localization in images containing only one face on a relatively uniform background within light variations. For that we proposed a robust colour segmentation approach in the TLS space which uses a modified watershed algorithm. To extract the facial features (like eyes and stops), we combined a kmeans clustering method with a geometrical approach and applied it on the segmented region of the face. We also proposed a 2D/3D multimodal approach which uses a weighted fusion of the scores obtained by the modular “EigenFace” and our 3D anthropometric facial signature. We evaluated our 3D and 2D/3D face recognition approaches on a sub base of IV2 which contains stereoscopic images of several human faces. The obtained results are very interesting compared to classical techniques of 2D face recognition. Finally, we discussed how to improve the performances of the proposed approaches.

No description available.

No description available.

Pain and performance anxiety (PA) are common problems among string players. Therefore, the purpose of this study was to assess and compare PA and prevalence rates and locations of pain in violinists, violists, cellists, and bassists. Subjects completed a questionnaire that included sections on demographics, musical background, practice habits, musculoskeletal problems, non-musculoskeletal problems, and PA. Anthropometric data were gathered on all 115 subjects. Results show that there are differences in both pain and PA across instrument groups. Violinists reported the highest number of pain sites, followed by violists, bassists, and cellists. The left shoulder was the most-often reported pain site, followed by the neck and right shoulder. Aching was the most cited term selected to describe pain. Several anthropometric indices were significantly correlated with pain, notably right thumb to index finger span in both cellists and bassists. In all instrument groups, at least one pain site was significantly correlated with one of four PA questions. Results warrant the development of intervention strategies and further study of the relationship between pain and performance anxiety.

Over nine million children 6-17 years old are overweight in the United States. They may face the highest risk ever for overweight, obesity, and associated health problems as adults. This study’s purpose was to evaluate adiposity measurement methods in children by comparing accuracy, cost effectiveness, ease of use, and practicality. IRB approval, parental consent and student assent were obtained. Data were collected by trained employees at six elementary schools on 789 subjects for boys (n= 391) and girls (n=398) 5-11 years old African American children. Hip and waist circumferences, height, and weight were collected. BMI, Waist to Hip Ratio (WHR), and Waist to Height Ratio (WHTR) were calculated. Upon analysis, weight and hip circumference were correlated (r = .94135) and waist circumference was correlated to BMI (r = .92396). The measures of adiposity were highly correlated indicating that a variety of anthropometric measurements can be used for measuring children.

No description available.

Knowledge of anatomy is an elementary step towards the understanding of the human body. First used by Alphonse Bertillon as an identification system, anthropometry refers to the measurements of human individuals. In orthopaedics, comparative analysis is widely used in the understanding of morphological variance due to races, sex and pathological conditions. The characterization of bone and joint geometry has also been a foundation of modern surgical implant design. Traditional anthropometric studies rely on physical measurements by means of osteometric table. Recent advancements of 3-D imaging modalities and image processing techniques have empowered more fine-grained anthropometric characterization. The inspiration for the study is: - the understanding of anatomy originating from the clinical domain have shown to contribute to undesirable inconsistency in the image processing domain. - the difficulty of existing automated anthropometric methodology in handling pathological femur. - the tedious amount of manual and subjective work involved with the increasing amount of high resolution imaging data. The aim of the study is to: - develop a consistent and robust methodology in accurate extraction of anthropometric parameters on the femur. - increase the level of automation on the process of anthropometric parameter extraction. With the bridging of anthropometry and the image processing disciplines, a robust methodology of anthropometric parameter extraction with high level of automation was developed, implemented and tested. A dataset comprised of femoral CT scans of 19 healthy Australian, 10 healthy Japanese, 15 Japanese diagnosed with primary or secondary hip osteoarthritis and 20 adult sheep was utilized for testing. Intra-class correlation and Cronbach's α were extensively employed to evaluate the intra-rater, interrater and repeated scans consistency of the proposed methodology. High correlation values (mean > 0.95) were noted suggesting a high consistency of the methodology. All healthy and osteoarthritis human datasets were processed successfully. With the structural similarity between the sheep and human femur, the robustness was further demonstrated by accurate processing of the sheep dataset without the need of any modification of the underlying methodology. The methodology proposed is highly automated and requires very few user interactions in the parameter extraction stage.

No description available.

No description available.

The overall objective of this thesis is to compare and contrast alternative measures of wellbeing in New Zealand during the 19th -early 20th centuries from a spatial perspective by collecting, collating and analyzing new economic, social and anthropometric data. Provincial data was collected from the Statistics New Zealand Annual Reports and New Zealand Census. Anthropometric data was derived from the personnel records of New Zealanders serving in WWI, which only became available to the public in 2005. Time-series tests for convergence and causality have been applied to analyze New Zealand’s economic history, where appropriate. The last quarter of the 19th century in New Zealand was a period of rapid change both in terms of economic and demographic indicators. Prior to the universal convergence of the existing monetary-based measures of wellbeing across Provinces, there were some apparent disparities in the commodity price and real wage series, as well as urban-rural differences in occupation-specific real wages and infant mortality trends. There was also no single pattern of stature decline across provinces during 1871-1898, or between urban and rural areas, where disparities were particularly apparent. The traditional view of the healthy and wealthy New Zealand could only be established at an aggregate level, during a certain time period and for a certain ethnic group (New Zealand European only). Using Provincial data for the period 1874-1919 I have been able to show that improvements in real wages and a decrease in education inequality (between females and males) corresponded to lower infant deaths and thus better health outcomes, while increased dwelling density created unfavorable conditions for infants’ chances of survival. Anthropometric data was used in conjunction with socio-economic provincial data to establish the relationship between stature, urbanization, real wages and infant mortality. The results showed that dwelling density (overcrowding) and general economic conditions were both important in determining stature outcomes during 1870-1900, while the effect of infant mortality on stature was negligible. Most importantly, it has been demonstrated that in New Zealand stature represents a much more robust measure of living standards than real wages or health indicators on their own, at least during the 1870-1900 period.