Shape processing for digital anthropometryM. Annichini, R. Arena , M. Fanini, M. Fattorel, D. Pavei, D.Tasson, V.Garro, C. Lovato and A.Giachetti

Dept. Computer Science, University of Verona, Verona, Italy andrea.giachetti@univr.it

Recent advances on scanning techniques make possible toacquire high resolution models of the human body that can beextremely useful in order to assist people involved inanthropometric studies, but also for other applications likemedical diagnosis, clothing design, computer animation andentertainment.Our group, in collaboration with the Department of Motor andSport Sciences of University of Verona, focuses on theacquisition and analysis of human body shape, trying toconnect the science of anthropometry with the fields of shapeanalysis and mesh processing. This approach seems to offer amutual scientific enrichment to the involved disciplines.Traditional anthropometry largely relies on tape and calipersmesaurements, that are generally limited to 1D information.Body scanning technology provides 3D data of body surface,extending measurement capabilities to complex geometricalfeatures, body volumes and surface areas.

Human analyzer softwareWe developed a pipeline for digital anthropometry that, startingfrom body scanner data, automatically pre-processes them andperforms measurements. Pre-processing is realized usingVCG-Meshlab scripts and removes noise and holes, remeshingthe model. The following steps are the following:

• the skeletal tree of the shape is extracted using a methodbased on voxel coding and active contours

• trunk and limbs are roughly segmented on it

•a stick figure is initialized from this segmentation andsubsequently refined

•anthropometric measurements are extracted on the segmentedstick and surface

Health-related applications of this pipeline are described in[LCF*09] and [LMP*11]

Figure 1: Human analyzer pipeline

Web based interactive visualization

Figure 2: WebGL visualization of annotated human scan.

We implemented a WebGL application prototype able to loadand display 3D body models with associated data, i.e. the stickfigure as computed by the Human Analyzer software and a setof annotated 3D points. These points can indifferently beanthropometric landmarks or automatically computed salientpoints of the shape. The application lets the user navigatearound the 3D body and show/hide interactively differentstructures. Measurement and shape comparison features will beadded in the future.

Kinect-based anthropometric applicationsWe are developing anthropometric acquisition systems basedon the Kinect sensor. Fig3 shows a simple body measurementtool composed by a PC and a single device positioned in frontof the subject. The interface suggests the correct pose to thesubject, giving a feedback on the correct positioning of thelimbs and acquire and processes point clouds showing themeasurements set.

Figure 3: A kinect-based human body measurement system.

DiscussionApplications of human body shape analysis to data derivingfrom cheap 3D sensor have a real possibility to improve thepublic health in a society where obesity and related metabolicand cardiovascular risks are pandemic. The rapidly expandingfield that our team is exploring seems to offer concreteopportunities to apply scientific knowledge in various aspectsof everyday life allowing the development of applications thatcan be useful and commercially interesting.

References[LCF*09] LOVATO C., CASTELLANI U., FANTONI S., MILANESE C.,ZANCANARO C., GIACHETTI A.: Computer assisted estimation of anthropometricparameters from whole body scanner data. In Modelling the Physiological Human,Magnenat-Thalmann N., (Ed.), vol. 5903 of Lecture Notes in Computer Science.Springer Berlin Heidelberg, 2009, pp. 71-83.

[LMP*11] LOVATO C., MILANESE C., PISCITELLI F., ZANCANARO C.,GIACHETTI A.: Health-related shape analysis of 3d body scanner data. In 2ndInternational Conference on 3D Body Scanning Technologies (2011), pp. 87-94.