[FrontPage Include Component] Home

NAVIGATOR: Back - Home > Adi > Services > Support > Manuals > Apas > Dos :



Frame Grabbing


Site Map
About us


Contact us

General info

+1 949 858 4216

Sales & support

+1 619 992 3089






[FrontPage Include Component]


5.0 CAMERAS   Recording film or video sequences is not difficult, however, certain rules must be followed to ensure that subsequent analysis will be correct and accurate.

If three-dimensional movement analysis is desired, at least two cameras must be used. Additional cameras, up to a total of nine, may be used. Experience has shown that two or three well placed cameras are usually adequate to record most types of human motion.

The cameras must be placed so that each one can record the entire sequence of motion performed by the subject. Avoid camera placements that obscure important body joints during critical phases of the activity (i.e. a rear view of a golf swing would obscure the arms, hands and club near impact). At the same time, try to fill as much of the frame with subject activity as possible to ensure maximum resolution when viewing this recording.

Finally, orient the cameras so that their viewing planes are a right angles, approximately 90 degrees from one another. A ten or twenty degree variation from this is not critical. However, if the viewing planes differ by less than 30 degrees, the accuracy of measurement will be significantly reduced.

CALIBRATING IMAGE SPACE   In order to calibrate the image space, the area where the subject is being recorded and the location of a fixed point must be determined. These are the control points. Two-dimensional analysis requires a minimum of two points, though at least four co-planar points are preferred using the Direct Linear Transformation. For a three-dimensional analysis, at least six control points must be available with the additional requirement that all do not lie in the same plane. Six points marked on the floor or wall could not be chosen. The points should be spread to cover the entire space volume from low to high, side to side and front to back. The system will accept as many as thirty control points, but six to twelve points are usually adequate for an accurate calibration. The control points must be clearly visible from each of the cameras and the area enclosed by these points should approximate the image space. In other words, the image space should not be several times larger or smaller than the control points.

The control point fixture should be approximately the same size as the event being filmed. For example, if a full body gait is being filmed the control point fixture should be about the size of the person. However, if hand movements are being filmed a small control point fixture is recommended.

It is not necessary for the control points to be seen during the filming. It is common to use an object of known dimensions, such as a large tripod or cube-shaped frame (Figure 5-1), with clearly marked points for the control points. This object is placed in the area that will be filmed and then removed. In this manner the control points will not interfere with the activity being filmed. It is recommended that the control points be filmed before and after the activity is performed. This allows two shots of the control points, with one as a backup if something happens to other one. The camera field of view must be fixed for the duration of each motion sequence recorded for analysis.

Unless the Ariel Panorama hardware is utilized, the cameras must not be moved, panned, or zoomed during filming, either during the activity or between recording the activity and the control points.
Control point locations are measured in a Cartesian coordinate system. A sample cube measurement may be found in Appendix C, Sample Measurements and Orientation for Control Points. Each point location has an X, Y, and Z coordinate value. The origin of the coordinate system and the orientation of the coordinate axes are completely arbitrary (provided that X, Y, and Z are orthogonal and form a right-handed system). However, a wise choice of coordinate directions will aid in viewing and understanding the results of analysis. Typically, the Y coordinate axis is chosen to coincide with the vertical direction, while X and Z is often chosen to coincide with some significant direction associated with the activity.

If a large tripod or frame is used for the control points, one corner is usually designated as the origin and the balance of the points are measured relative to this point. This object is then oriented in the activity space so that the coordinate axes point in the desired directions.

5.2 RECORDING AN ACTIVITY   When filming an activity it is essential that all cameras record the event simultaneously. A subject should not be filmed from one angle and then again from another. These are two separate activities and will not give the same data.

The cameras do not have to be synchronized. Each camera does not have to record the same frame at the same time. Cameras do not even have to record at the same rate as long as filming speed of each camera is accurately known.
The one requirement that must be met in order to combine simultaneous camera view is that all cameras must record a single distinct event called the synchronizing event. Examples of such an event might be the impact of a falling object, a flash of light, the beginning movement of a golf ball, tennis ball, or clapping hands. Also a fixed point must be in the view when filming the activity and the control points. This is discussed later in
Chapter 7: Digitizing Module.

Some final suggestions regarding film and video recording are as follows:
bulletUse quality cameras and quality lenses with proper focus and adequate lighting. A subject's motion cannot be digitized if it cannot be seen.
bulletMatch the camera speed to the activity under study. Normal human motion can be analyzed using 30 frame/second video. High speed activities, especially those involving any type of impact, must be filmed at higher frame rates, often 200 frames/second or more.
bulletIt may be necessary to experiment with filming to determine the optimum frame rate for certain activities.
bulletThe use of higher shutter speed will record clearer image and minimize the blur of a fast motion. However, at high shutter speeds there may be a significant loss of exposure light.
bulletWide angle lenses should be considered when filming in close areas.
bulletAdditional lighting may be necessary.
5.3 FILMING - QUICK REFERENCE   Before any filming occurs the following question should be asked.

  1. How long is the sequence?
  2. What are the significant points?
  3. Is the motion fast or slow?
  4. What are the lighting conditions?
  5. How much space is there?
  6. Can the joints be marked effectively?
  7. How many trials should be filmed?
  8. Is the background cluttered?
  9. What is the synchronizing event (internal or external)?
  10. How is the subject information recorded?
Once these questions and any other concerns are answered then data collection can begin.
  1. Set cameras so that the motion can be seen clearly (they should be at least 30� apart and no greater than 150�).
  2. Select appropriate shutter speed:
    bulletwalking - open shutter
    bulletjogging/sprinting - 1/250 to 1/500
    bulletgolf/baseball - 1/1000
  3. Get additional lights if necessary.
  4. Determine synchronizing event.
  5. Focus and frame image. Make certain cameras are set to Manual focus.
  6. Check batteries or external power supply.
  7. Record control points.
  8. Mark trial.
  9. Record event.
  10. Record control points again.

Up Next

This page was last modified on 12/07/2008 at 22:47 PST. Copyright � 1994 - 2002, all rights reserved, Ariel Dynamics Inc. Please send your comments or feedback to or proceed to our feedback form. This page has been accessed many times since Dec 12, 2002. Our privacy policy is here.