From Chapter 3:


      Original digitization using manual clicking

      Drawbacks- 1) amount of time required

                  2) human error of digitizing

      200 Hz= 7 hours of digitizing- This is what led historically to smaller subject

      numbers for biomechanical motion analysis.


Use of 2 dimensional versus 3 dimensional analysis-

2 dimensional for more planar movements

3 dimensional for more rotary techniques. However still need to use at least 2 cameras to

obtain the original data and then reconstruct a 3 dimensional image from the two cameras

worth of data.


Applications for calculating movement parameters-

Can either use the calculations from a computer- are then limited to their imaginations

can write software to calculate what you want for the sport that you want.

The current software language used is C++ because it is modular and users can combine

programs. The programs can be complete within themselves ( not unlike subroutines) and

then be called up by another computer. This is good group programming software. This is

a very marketable skill to have if you want to move into industry using computers.

Also have software with some systems to integrate the video with force plate data via

synchonizing the signal electronically


Hardware development-

Video cameras- much cheaper to use than high speed cameras but somewhat less power

due to slower sampling rate.

Electronic versus mechanical shuttering- In either case it is the difference between how

many samples are taken per second versus how long the shutter is open for each particular

frame. If the shutter is open for too long get blur of the image.

High speed video from regular 30 Hz video is obtained by de-interlacing the video lines of

the image.can introduce errors but you want a faster frame rate so you hope the filtering

will remove the data introduced by this sytem.

or you can a system that never interlaces in the first place and saves that data.


Split Image High-Speed Video

Electronics scans the top 1/3 at a rate of 180 hz.

( 3 X 60Hz) but for only a 1/3 of the picture.

So there is not additional information being sent to the camera.

It's not the speed that's limiting it's the information rate and that stays the same.


High speed cameras-

Digital cameras are generally limited by the information rate going to the recorder- The

higher the frame rate and the better the resolution, the higher the information rate being sent

to the computer.

Most cameras are analogue. Newer cameras are digitial from the still picture format, and

even hand held video cameras are now digital. You can scan through and get any digital

image you want.

High speed recorder / output units

These cameras actually have the shutter opening and closing at a faster frame rate much

faster then the usual video recorder.

Television cameras that record at a 100 Hz or faster for sporting events record at actual

faster frame rates.

They have to have much more finely tuned mechanical parts and faster rates of acceptance

to the playback unit.

Compressed video and digital images

Mpeg- motion picture experts group

Jpeg- joint photographic experts group

These are the same type of compression that are used when sending images across the

internet. Tend to lose information while compressing.

Video processing- Marker tracking methods

      retroreflective markers-

different sized targets- so computer looks for different sizes for different markers

colored targets- Actual colors or black and white.

Finding the centroids of certain markers and tracking them over space by predicting where

they will for the next frame.


brightnes of markers- different brighnesses for each marker


In general with x,y,z coordinate data we filter the x and y , z coordinate data separately.

Often there is more movement in one plane than the other and it is also faster to filter them

separately so that is how it is done.

For video data are generally filtering out the higher frequencies and are leaving the slower

lower frequencies that represent movement. The assumption is that the lower frequencies

mostly represent movement and don't represent error.

The Butterworth filter we use is a 5 point moving average that emphasizes the middle point

the most and emphasizes the surrounding 2 points on either end less as they move out from

the main point of interest.