Navigation, Visualization and Searching in Databases of Full Body 3D Scans
Marc Rioux
Authors: Marc Rioux, Eric Paquet, and Zouhour Benazouz
Institute for Information Technology,Visual Information Technology,
National Research Council of
Three-dimensional raw data of the human shape consist of 100s of thousands of coordinates. Similarly to traditional anthropometry, 3-D anthropometry will have an impact on industrial design when data reduction is applied to the raw data. The objective is to remove redundancy in the data set and, extract the most compact representation that is relevant to the understanding of human shape variability. Two cases of data reduction will be presented here.
The first is an approach based on compact shape descriptors to automatically index large dataset of human shapes. A short review of digitalization of human bodies is presented. Then it is shown how it is possible to describe the three-dimensional shape of the bodies by representing them with compact support feature vectors. A recurrent data mining system based on these vectors is then presented. This technique allows us to search and navigate the database in real time.
The second case essentially removes human shape data redundancy using principal component analysis on the geometry of the 3-D raw data. This is a method for extracting main modes of variations of the human shape from a 3-D anthropometric database. Previous approaches rely on anatomical landmarks. Using a volumetric representation, we show that human shape analysis can be performed despite the lack of such information. This approach allows us to visualize and understand the human shape variability along few relevant components which body shape variability.
Shape descriptors reduce the human shape data set to few 100s bytes per subject. Principal components reduce it down to few 10s of bytes. Applications of both methods will be presented in the context of engineering anthropometry.