View Full Version : Re: sampling rate

Louisa Sh Law [rs]
11-13-2002, 07:52 PM
Dear all

Thanks for the replies on my concerns on sampling rate. In general, most of the replies indicate that 60Hz of sampling rate should be sufficient.
The following are all the replies for your review
Louisa Law

I think it depends on how involved the cp kids are - if their upper limbs
are involved and they therefore have jerky movements you might need a
higher sampling rate to capture this accurately.

Your sampling rate is more than adequate. I used 60 Hz to examine
volleyball spiking mechanics and, in a post-hoc Fourier analysis, found
meaningful signal power in frequencies up to 10-15 Hz (max).

Peter F. Vint, Ph.D.
Research Scientist
Research Integrations, Inc.
9280 S. Kyrene Rd. Suite 101
Tempe, AZ 85284-2952
Phone: (480) 893-1600 x214
Fax: (480) 893-0602
e-mail: peter.vint@riimail.com
URL: http://www.researchintegrations.com

The answer depends heavily on the expected velocity of the throw. For example, if the children will throw the ball in a pattern similar to that used when throwing a dart, then you may get by at 60hz. If they're throwing more like a baseball pitch, then 60hz will likely give you only a handful of frames (maybe 5?) to describe the acceleration phase of the arm motion. We record college level pitchers at 240 frames/sec and collect approximately 10 frames for the acceleration phase, and I'd be reluctant to do the inverse dynamic calculations with fewer frames.


I think 60Hz will be enough, when you are
not looking at speed. However, it is difficult to track all markers in these
CP-children, as they rotate a lot with their body and their hands are
obviously not in a straight line with their arm. So you need quit a lot of
cameras to make sure you don't loose any markers. On top of that, the higher
the sampling rate the better, but 60Hz will do the trick!
Good luck and keep in touch?


Annieck Ricken
Postgraduate researcher
Manchester Metroplitan University
Centre for Clincial and Biophysical Research into Human Movement
e-mail: a.ricken@mmu.ac.uk
contact work: 0161-247 5504
contact home: 01244-318917

I doubt there is any point sampling faster than 60Hz. 100Hz is probably
a convenient number and it may have to match other equipment;
you can justify the lower sample rate by considering the velocities and
accelerations at higher frequencies in the
case where the motion is not linear/constant. I use low pass filters for
digitising accelerations
if I sample at, say 60Hz but I don't need to worry with optical
measurements since I know
that signals above noise are physically impossible at high frequencies.
This is my experience with structures
But it carries over to human dynamics I should think.

James Brownjohn

I performed 3D kinematic analysis on a specific field hockey skill (push-in)
and utilised a sampling rate of 50Hz with a 1/1000th second shutter speed. This
was quite sufficient for the movement being recorded and was able to track a
ball with a velocity of up to 20m/s.
Most papers that I have encountered measuring faster movements such as the
hockey ball trajectory after being hit (ball velocities over 30m/s) have
utilised the faster 100Hz sampling.

Rebecca Kerr B.SpExSc (hons 1) MAAESS
PhD Candidate
Technical Officer
Institute of Sport and Exercise Science
James Cook University
Townsville Queensland 4811

Phone: 07 4781 4952
Fax: 07 4781 6688
Email: rebecca.kerr@jcu.edu.au

According to Shannon's Theorem, which states that: " .. if the sampling is carried out at more than two times
the highest frequency in the signal before sampling, then the signal can be reconstructed without degradation.",

your minimum sampling frequency (pictures per second) should be twice the highest frequency component of the measured motion.

I hope this information is useful.

Good luck


A. Univ. Prof. Dipl. Ing. Dr. Christian Peham
Clinic of Orthopaedics in Ungulates
University of Veterinary Medicine Vienna
A-1210 Vienna
Veterinaerplatz 1
Phone: ++43/1/250 77-5506
Fax: ++43/1/250 77-5590

frequencies of normal, fast movement are around a couple of Hz (3-7) (finger
tapping - hand tapping). Anything above 10 Hz are very fast tremors.
Our customers typically measure movement of upper limbs with sampling rates
between 25-60 Hz.

The sampling theorem has to be understood as following: a two times higher
sampling rate (than the signal) is sufficient to reconstruct the frequency,
amplitude (and phase) of a sinus like signal. So if you are interested in
the frequency of hand tapping it would be sufficient to sample at about 15
But what you are going to measure is different form a sinus like signal and
you may be interested in the detail of the movement ('the curve') and the
inter differences between movement trials.

I think 60 Hz should be a good sampling rate if you do not mind missing very
small details.

Mit freundlichen Gruessen / Best regards,

Reiner Beck
Dipl. Ing. (FH)

zebris Medizintechnik GmbH
Hintere Grabenstrasse 26-30
D-72070 Tuebingen
Tel +49 7071 27003
Fax +49 7071 27005
e-mail: beck@zebris.de
Internet: www.zebris.de

It has to be around 12Hz which coincide with the fastest tapping movement
one can do with his finger on a table.

Jo Van Vaerenbergh

it is not so hard to go faster with technology these days, if you are
talking about instrumentation and not video, why not sample at 1000 hz or
more? for dummy movements in car crashes, the sampling rates are 1000 hz or
more, always. actually, most of that stuff is 10 kHz. it is then hardware
filtered to around 3000 hz and then software filtered depending on the data
for loads, accelerations, etc to cfc (channel filter class) 180, 600, or
1000. we hardware filter to 600 hz cutoff all upper extremity stuff. then
you can filter as much as you want to afterwards. if you are talking about
video, then it gets more expensive, and it depends on the speed of the
event... we film stuff at 1000 fps but this is for impacts. if you need more
info i can send you a paper we wrote for the upper extremity. most of this
stuff is impact-related, i am not sure if it applies for limb movement for
what you are doing...

The power spectrum of human movement in general (including limb
movements) is largely contained