Chapter 3 Instrumentation
3.5 Software and data processing
Data acquisition and data post processing were performed in Vicon Nexus 1.4. The aim of the data post processing was to:
1) Compute the 3D locations of each marker in each time frame using the raw video data from the MX cameras (reconstruction).
2) Label each marker according to the marker model (section 3.5.2).
3) Export the computed locations and trajectories (kinematic data) and other data such as forces and EMG data for further analysis. In this thesis, further analysis of kinematic data was performed in Matlab 2009b (MathWorks®, 2013).
The reconstruction and labelling processes are shown in Figure 3.5 and described in section 3.5.1 and 3.5.2.
Figure 3.5a. Images from three of the twelve MX cameras after intensity threshold to identify the markers. The frames of the camera sensors are shown by white rectangles; the markers are
shown as white dots and dark blue dots are masked area where data are ignored. b. Markers reconstructed into a 3D digital space, calibrated to a floor shown by a white grid. The force plate is shown by a grey square labelled 1. A red arrow shows the ground reaction force at the
force plate, representing its origin (CoP), magnitude and direction. The camera frames have been calibrated against lens distortion. c. Each marker is labelled according to the Vicon Plug-in
Gait marker model. The markers are linked to other markers according to body segments. d. Shows a 3D perspective view, the ground reaction force is clearly visible against the left foot where the foot is in contact with the floor during walking. e. A DV camera view with an overlay
3.5.1 Kinematic data reconstruction
The MX cameras recorded video data from different angles and detected the IR light reflected from the markers; the video footages are thresholded to exclusively detect the markers. To reconstruct the markers’ locations in a 3D space, Vicon Nexus applies multiple view geometry computation (Hartley, 1999, Hartley and Zisserman, 2004) to compute and provide the 3D position of each marker from the 2D marker positions in the camera planes. The system was calibrated using the recommended method by Vicon. The Vicon MX system used in this work has a reported precision of 1mm and accuracy of ±0.1mm. The measured linear precision error was 0.4mm and the measured gradient error of the x-y plane in the reconstructed space against the horizontal plane in the measured space was less than ±1°. These errors were obtained by placing 8 markers on the floor distributed evenly on a ring of circle with 1m radius from the origin. Two markers were aligned along the x axis and two markers were aligned along the y axis. Reconstructed locations were compared against measured locations. The markers aligned with the x and y axes were used to compute the angular error and the remaining four markers were used to ensure the reconstructed marker locations were on the same plane.
3.5.2 Marker labelling
The markers in the 3D space were labelled according to a marker model, such as the Plug-In Gait marker model (Vicon®, 2010). An example of a motion capture trial of the walking motion using the Vicon Plug-in Gait model is shown in Figure 3.6. A marker model is the layout of markers used in motion capture, in the Vicon Plug-in Gait and the arm marker model used in this work, the markers are placed in anatomical positions so that the location of body segments or joints can be located. All the defined markers in the marker models have unique names. In most instances, several markers are associated together to form a segment in the marker model. For example, in Figure 3.6, four markers around the head of the subject are associated together to form the head segment of the model, represented by a white block in the 3D reconstruction. Markers can also
be part of more than one segment, for example a marker placed next to the knee can be associated with the segment representing the upper leg and the segment representing the lower leg. The relative orientations between segments can be used to calculate joint angles, for example knee joint angle. A detailed description of the marker model of the arm used in this thesis was developed to measure movement of the arm and is described section 5.3.
Figure 3.6. The human gait in digital form. Markers are attached to the subjects at different locations, the trajectories of these markers are shown in yellow in the left image and in blue in the right image. The marker location in real space is reconstructed in a 3D space shown in the right image (with perspective), the markers had been associated to body segments such as the upper leg, and form different rigid body segments.
3.5.3 Export data for analysis
After the markers had been reconstructed and labelled, the trajectories of the markers, force data and EMG data were exported in the form of comma separated variables (CSV) files. The CSV files are imported into Matlab® 2009b for further analysis.