CAPÍTULO II: MARCO TEÓRICO
2.4 LA BIBLIOTECA
2.4.4 Sistemas de organización
Gait Laboratory contains the Vicon system, a video-based optoelectronic sys- tem (Oxford Metrics Group, 2007). The capture room and camera layout are illustrated in Figures 4.2a and b, respectively. There are 12 Vicon cameras, MX1 to MX12, and one force plate, model OR6-7 (AMTI, 2007). The sur- face of the force platform is at the same elevation with the surrounding floor tiles, leaving a horizontal gap of 3 mm. This set-up aims to prevent occur- rence of both mental and physical obstructions that may arise while the TS is
performing the stamping activity during an experiment.
The MCS consists of high speed and low latency cameras to allow for capturing real-time motion. Each camera records 200 two–dimensional (2D) frames in every second. The cameras are equipped with wide angle lenses. The aperture setting (in the range of 1.4–16) is set at 2.8 to control the amount of light passing through each lens. This parameter also stipulates the depth of field, i.e. the range of distances from objects to the camera lens that remains in sharpness, a requirement for the system to recognise markers. The capture volume, also known as the field of view, during experiments described in this chapter is 2×3×2.2m (width, length, and height, respectively, indicated as
Figure 4.2: Gait Laboratory at the University of Warwick: (a)capture
room, (b)camera layout, (c) camera with a strobe unit, (d) view of a
shaded area in plan in Figure 4.2b).
The Vicon system uses passive markers, i.e. markers that reflect the light. The system is different from the alternative active markers system, in which markers emit the light. The advantage of the passive markers is that the TS is not required to wear wires and batteries to power the markers during trials. The use of the system, however, still faces challenges. Since the markers do not have unique identifiers, the anatomical landmarks on which they are placed are not recognised automatically. The post-processing of the acquired data requires manual labelling of each marker in Vicon Nexus, a specialised software used for capturing and analysing the data (Oxford Metrics Group, 2008). This step entails assigning each marker on the TS to a corresponding
point on the body. This operation results in creating a TS’s body model
within the software, which can be used as a marker template. To apply this predefined template to a different TS, a procedure named “Subject calibration process” is employed (Oxford Metrics Group, 2008). In this step, all markers are captured while the TS is in a stationary (standing) posture. Afterwards, markers are labelled and the marker layout representing the body shape of the particular TS is fitted to the model template. Moreover, distances between labelled markers are measured and can be used later for automatic labelling.
The marker used in this study is 14 mm in diameter and is coated with a highly retro-reflective material. Each marker has the negligible mass of about 2 grams. To illuminate markers, the front of each Vicon camera contains a strobe unit (Figure 4.2c) configured with light-emitting diodes. When markers are visible inside the view range of a specific camera, rays of light from the strobe hit the markers and are reflected back to the camera lens. To optimize
the system’s performance, each lens is fitted with an optical filter so that only the reflected light having similar wavelength with the one generated by the strobe unit can pass into the camera lens. The image from each camera is processed by the system, in which centroid-fitting algorithms are used to determine which objects are most likely to represent markers (Figure 4.2d). This process results in reconstructing the three–dimensional (3D) coordinates of markers within the capture space.
The light from light bulbs inside the capture room does not interfere with normal working conditions of the system. Any strong light sources have to be removed from the capture room (e.g. the lab window is covered by curtains to prevent passage of sunlight, Figure 4.2a). Although, the Gait Laboratory is designed in accordance with recommendations from Vicon user manual, there are still non-marker reflection points inside the capture volume. To avoid the possibility that the system might recognise them as markers, these points are removed within the software by a process called “Masking MX Cameras” (Oxford Metrics Group, 2008). The 3D space inside the capture volume is established through a dynamic calibration process. The system records the movement of a calibration wand with five fixed marker points until a sufficient number of frames are captured by all cameras. Based on these reference frames and known geometry of the wand, the system calculates the relative distances between cameras and their projections. By placing the wand on the floor and levelling it (Figure 4.2e), the horizontal plane of the 3D space is defined by the two perpendicular axes of the wand. After this step, the system is capable of using data from multiple cameras to reconstruct the 3D space inside the capture volume (Hasan et al., 1996). If a marker can be recognised by at least
two cameras, its spatial coordinates can be determined.
In static condition, the background noise of spatial data can be checked by monitoring stationary markers located within the capture volume for 30 s (Hasan et al., 1996). In the Gait Laboratory, the maximum 1 s RMS of the background noise was found to be less than 0.05 mm, which is considered acceptable for the current study.