Table 1.
Participants recruited for this study.
Fig 1.
Experimental setup used for collecting 3D body scans.
We manually revolved the Kinect V2 camera around the participant following the scanning perimeter. Participants stood with their arms abducted on a transparent platform.
Fig 2.
We aligned imported 3D body scans such that body segment segmentation along the frontal plane was perpendicular to this plane.
We aligned the imported 3D scans to the MeshLab global coordinate system as shown.
Fig 3.
From 3D scan to BSP output estimations.
1. A representative 3D scan in MeshLab. 2. We segmented each 3D scan into 16 individual body segments. 3. Segmental coordinate system definitions for each body segment showing anteroposterior (red), mediolateral (green), longitudinal (blue) axes. Here the red dot is the center of mass. We aligned the foot segment coordinate system such that the longitudinal axe was along the long length of the foot.
Fig 4.
After placing anatomical landmarks on the participant, we used a Microsoft Kinect V2 and Microsoft Builder 3D to collect repeated 3D body scans. We exported the acquired scans into MeshLab for removing extraneous data, globally aligning, and segmenting each scan into 16 body segments. We then wrote custom MATLAB scripts to estimate the BSPs of each body segment.
Fig 5.
Total body volume estimated with our proposed 3D method for both males and females.
Comparison volumes between the three scans (Scan A-C) are shown for each participant (coloured dot). The box is the 95% confidence interval around the mean (red line) and SD about the mean is shown (vertical blue line).
Fig 6.
Body mass (kg) estimated with our proposed 3D method, the ECM method, and the medical scale for both males and females.
Each participant (coloured dot). The box is the 95% confidence interval around the mean (red line) and SD about the mean is shown (vertical blue line).
Fig 7.
Three different scans (Scan A, Scan B, Scan C) and three individual body segmentations from a representative male and female participant are shown. The colouring used for the segmented scan is meant to show clear segmentation borders between adjacent body segments.
Table 2.
The mean coefficients of variations (%) for repeated body segmentation of repeated 3D scans.
Table 3.
Body segment mass estimates (mean±SD in kg) obtained using the proposed method, the ECM approach, and regression modelling for both male and female participants.
Table 4.
Moment of inertia estimates (mean±SD in kg cm2) obtained using the proposed method, the ECM approach, and regression modelling for both male and female participants.
Table 5.
Longitudinal length estimates (mean±SD in cm) obtained using the proposed method and the ECM approach for both male and female participants.
Table 6.
Center of mass position from the proximal endpoints (pCOM) (mean±SD in %) estimated using the proposed method and the ECM approach for both male and female participants.
Fig 8.
Top view of full-body, head and hand scans.
1. An example showing lack of texture in the scan on the upper arm regions. 2. Varied degree of detail in the head region with concave surfaces appears as filled in. Further to the right head suggests minor sway, as facial features are distorted more so than the head to the left. Hands showing varied degree of detail.