M. Rioux’s research while affiliated with National Research Council Canada and other places

Univariate anthropometric data have long documented a difference in head shape proportion between Chinese and Caucasian populations. This difference has made it impossible to create eyewear, helmets and facemasks that fit both groups well. However, it has been unknown to what extend and precisely how the two populations differ from each other in form. In this study, we applied geometric morphometrics to dense surface data to quantify and characterize the shape differences using a large data set from two recent 3D anthropometric surveys, one in North America and Europe, and one in China. The comparison showed the significant variations between head shapes of the two groups and results demonstrated that Chinese heads were rounder than Caucasian counterparts, with a flatter back and forehead. The quantitative measurements and analyses of these shape differences may be applied in many fields, including anthropometrics, product design, cranial surgery and cranial therapy.

Texture mapping is useful for constructing a three-dimensional (3D) model because a realistic D model can be obtained efficiently and quickly. This paper proposes a system to construct a D model using range intensity images. A range intensity image, which is also called a reflectance image, refers to the intensity image that is acquired simultaneously with the range image captured using an active range sensor. Such an image has an important property in which illumination conditions, such as geometrical arrangement and power of illumination, can be controlled at the capture time, which allows the estimation of the reflectance properties of the object. Several methods using range intensity images are improved and combined to construct an effective system, the registration of range images and color images is realized, an omni directional geometric model is constructed by registering and integrating multiple range images with range intensity images, and the influence of the illumination environment that occurs in color images is removed. In addition, a method to estimate the illumination color is introduced to compensate for the color of illumination light. Experiments show the effectiveness of the constructed system for obtaining a realistic 3D model.

Most active optical range sensors record, simultaneously with the range image, the amount of light reflected at each measured surface location: this information forms what is called a range intensity image, also known as a reflectance image. This paper proposes a method that uses this type of image for the correction of the color information of a textured 3D model. This color information is usually obtained from color images acquired using a digital camera. The lighting condition for the color images are usually not controlled, thus this color information may not be accurate. On the other hand, the illumination condition for the range intensity image is known since it is obtained from a controlled lighting and observation configuration, as required for the purpose of active optical range measurement. The paper describes a method for combining the two sources of information, towards the goal of compensating for a reference range intensity image is first obtained by considering factors such as sensor properties, or distance and relative surface orientation of the measured surface. The color image of the corresponding surface portion is then corrected using this reference range intensity image. A B-spline interpolation technique is applied to reduce the noise of range intensity images. Finally, a method for the estimation of the illumination color is applied to compensate for the light source color. Experiments show the effectiveness of the correction method using range intensity images.

We study the behaviorally important task of gender classification based on the human body shape. We propose a new technique to classify by gender human bodies represented by possibly incomplete triangular meshes obtained using laser range scanners. The classification algorithm is invariant of the posture of the human body. Geodesic distances on the mesh are used for classification. Our results indicate that the geodesic distances between the chest and the wrists and the geodesic distances between the lower back and the face are the most important ones for gender classification. The classification is shown to perform well for different postures of the human subjects. We model the geodesic distance distributions as Gaussian distributions and compute the quality of the classification for three standard methods in pattern recognition: linear discriminant functions, Bayesian discriminant functions, and support vector machines. All of the experiments yield high classification accuracy. For instance, when support vector machines are used, the classification accuracy is at least 93% for all of our experiments. This shows that geodesic distances are suitable to discriminate humans by gender.

The Karhunen-Loève expansion is used for a compact 3D description of the human body shape. In this paper we show that a compact 3D description can be achieved using a small set of eigenvectors. Experimental results using a database of three-dimensional body scans are presented. Applications are also discussed. The measurement of the human body (Anthropometry) is an essential part of the engineering design of cars, aircraft, workspaces, and clothing, to name a few. Traditionally such measurements involve the linear distances between anatomic landmarks and the circumferential values at predefined locations. Practically, such measurements are limited to a set of about 100 values.

During the autumn of 2004, a team of 3D imaging scientists from the National Research Council of Canada (NRC) was invited to Paris to undertake the 3D scanning of Leonardo's most famous painting. The objective of this project was to 3D scan the Mona Lisa – obverse and reverse – in order to provide high-resolution 3D image data of the complete painting to help in the study of the structure and technique used by Leonardo. This paper describes some challenges associated with scanning the Mona Lisa and presents results of the modeling and analysis of the 3D data including preliminary measurements of the thickness of the varnish layer. Au cours de l'automne 2004, une équipe de scientifiques spécialisés en imagerie du Conseil national de recherches du Canada (CNRC) a été invitée à Paris afin d'entreprendre la numérisation 3D du tableau le plus célèbre de Léonard de Vinci. L'objectif de ce projet était de numériser en 3 dimensions l'avers et le revers de La Joconde afin d'obtenir des données d'image haute résolution en 3D du tableau complet afin d'aider à l'étude de la structure et de la technique appliquée par de Vinci. Cet article décrit quelques difficultés qui se sont présentées à la numérisation de La Joconde et il présente les résultats de la modélisation et de l'analyse des données 3D, notamment les mesures préliminaires de l'épaisseur de la couche de vernis.

Modeling and creation of realistic human heads have applications in various domains, including computer animation, electronic games, virtual environment, film production, etc. Often the characters designed by artists are far from being human-like. In crowd simulation, characters tend to have the same or similar face appearances. In 3D games, it is important for users to have plenty of selections of 3D avatars. We explore an approach, which makes use of 3D scans of real persons to reconstruct countless new faces. The reconstructed faces based on principal components analysis have differences in ethnicity, gender, face shape, height and weight. La modélisation et la création de têtes humaines réalistes ont des applications dans différents domaines, notamment l'animation informatique, les jeux électroniques, l'environnement virtuel, la production de films, etc. Souvent, les personnages dessinés par des artistes sont loin de ressembler à des humains. Dans les simulations de foules, les personnages ont tendance à avoir les mêmes visages ou des visages très semblables. Dans les jeux 3D, il est important que les utilisateurs disposent de nombreux choix d'avatars 3D. Nous explorons ici une approche qui fait appel à des numérisations 3D de personnes réelles pour reconstruire un nombre infini de nouveaux visages. Les visages reconstruits à partir d'une analyse en composantes principales présentent des différences sur plusieurs plans : caractères ethniques, sexe, forme du visage, grandeur et poids.

During the autumn of 2004, a team of 3D imaging scientists from the National Research Council of Canada (NRC) was invited to Paris to undertake the 3D scanning of Leonardo's most famous painting. The objective of this project was to scan the Mona Lisa – obverse and reverse – in order to provide high-resolution 3D image data of the complete painting to help in the study of the structure and technique used by Leonardo. Unlike any other painting scanned to date, the Mona Lisa presented a unique research and development challenge for 3D imaging. This paper describes this challenge and presents results of the modeling and analysis of the 3D and color data. À l'automne 2004, une équipe de spécialistes de l'imagerie 3D du Conseil national de recherches du Canada (NRC) a été invitée à Paris pour entreprendre la numérisation 3D de l'œuvre de Léonard de Vinci. Ce projet avait pour but de balayer au laser la Joconde (avers et revers) de manière à fournir une image numérique 3D à haute résolution de la toile complète pour faciliter l'étude de sa structure et de la technique employée par Léonard de Vinci. Plus que toute autre peinture soumise à un balayage numérique jusqu'à présent, la Joconde constituait un défi pour l'imagerie 3D, sur le plan de la recherche et du développement. Dans cet article, nous décrivons les difficultés rencontrées et nous présentons les résultats de la modélisation et de l'analyse des données 3D et des données sur les couleurs.