FAQ – frequently asked questions
Motion analysis describes the motion of bodies (e.g. human being or machine) with kinematic parameters like distance-time graphs, velocities and accelerations.
Thereby body motion is not evaluated as a whole, but the motion of single, selected parts of the bodies is described. This way, when evaluating human motion, the motion of joints is described, whereas, when analyzing machine motion, the motion of parts of the machine is tracked.
Basically there are two different analysis methods:
- On-line systems
- Off-line or video-based systems
Here active or passive markers are affixed to the body and in each case the position of the markers is saved. The passive markers are activated by radiation sources (infrared, ultrasonic) and the rays are reflected to a detector. The active markers emit rays to the detector by themselves.
The detector (camera) reflects only the position (by the ray) of every single marker and saves the x- and y-coordinate; images are not saved. These results are processed further mathematically and this way all kinematic parameters of motion are calculated and can be displayed.
The description of the motion on the PC happens quasi immediately after the real motion (i.e. on-line).
Off-line or video-based systems
All systems are based on video sequences (from standard or high-speed cameras), which are recorded of the motion. These sequences are digitized and stored on the hard disk of the PC. Afterwards (off-line) the evaluation is conducted by the help of image processing methods. Mostly there are also markers affixed (on the joints or machine parts) and the motion of the markers is then processed by the evaluation software.
WINanalyze belongs to the off-line systems. Above all it distinguishes itself by highly developed image processing methods, which allow a high precision and flexibility in tracking objects.
There is the possibility to track an object motion in an image sequence with subpixel accuracy. This means that motion can be determined to fractions of a pixel (up to 0.01 pixel), what results in a more exact analysis. Subpixel calculations are in a sense the same as extending the image resolution.
We use scientific methods of image processing to gain this subpixel accuracy.
The motion analysis software WINanalyze utilizes a combined method of interpolation, oversampling and a kind of error-correction, which is applied in the process of template matching. To do so a video sequence must be recorded according to the sampling theorem.
Scientific papers on this topic you can find on our WINanalyze Publications.
The knowledge of the coordinates of an object in at least two different image sequences and details on the positions and intrinsic parameters of all used cameras make it possible to calculate the real coordinates of the object motion.
This calculation makes it necessary to perform a calibration to establish a relation between camera coordinates (2D in pixel) and world coordinates (3D in mm). In WINanalyze the calibration is realized by the help of a mathematic camera model and a physical calibration model (with known world coordinate values). The calibration method according to Tsai (1986), which is implemented in the program, distinguishes itself by a high accuracy in the calibration results.
By using WINanalyze you can choose between four algorithms for the automatic tracking:
- SSD-Correlation (Sequential Similarity Detection)
SSD-Correlation is the most useful and fastest algorithm. It is useful for any kind of object.
Cross-Correlation is useful for sequences with changing lights.
- Texture tracking
Texture Tracking is useful for tracking markers.
- Manual tracking
Using the manual tracking you have to move to every object in every single image with the mouse and store the position by a mouseclick. For setting a point of reference you also take this algorithm, as by this means you obtain an absolutely fixed point without pixel jitter.
Using video-based motion analysis systems the analysis can be made automatically or manually.
When motion is tracked manually each marker/object must be evaluated from one frame to the next by a mouse-click (and the position must be saved).
When tracking is made automatically, the user only selects the marker/object in the first frame of the image sequence and then the software automatically detects all markers/objects in the rest of the image sequence. The user watches the process on the PC monitor and can interfere interactively.
Not only for the field of biomechanics, we have extended the motion analysis software with a body model module
– the object model – .
The object points in WINanalyze be associated with the object model. This allows to set very quickly whole groups of object points. In that case are calculated forces, moments and centers of gravity that occur when biomechanical movements.
Thus, among other e.g. in human movements predefined models used by Hanavan and Zatsiorsky. In addition, body models can also be defined by the user.