Dear subscribers,

Thank you for your responses and suggestions.
Further analysis of our videos revealed that the lights we used were out of phase, and therefore histogram equalization could not achieve the results we wanted. For the black and white videos, we used a temporal smoother in virtualdub that somewhat made the video nicer, but blurry, which could be a problem for quantitative results.

I copied my original post at the end of this message.


I have never dealt with this particular problem so I am not sure if this
would work, but in image analysis I have used the technique of histogram
equalization to improve the contrast of images (makes dark images
brighter and bright images darker). You can implement it in Matlab
using the histeq function.

Good luck!


I think the term you need to search on is colour normalisation/color
normalization. There was an IEEE conference paper called 'Colour
normalisation based on background information' that looked interesting but I
haven't read it and am only looking at the first few entries of a google


we have some experience with high-speed videos (10-12000 fps, see for instance found no easy way of removing the flickering without loosing information.

So, our solution was a combination of three factors:
1) No fluorescence light on in the lab
2) Ok for ambient incandescent light (no significant flickering)
3) High-intensitiy dedicated incandescence light on the test specimen

I hope this helps. Best regards,

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Luca Cristofolini, PhD tel: 39-051-6366864 (operator)
Lab. di Tecnologia Medica tel: +39-051-6366577 (direct)
Istituto Ortopedico Rizzoli fax: +39-051-6366863
Via di Barbiano 1/10
40136 Bologna, Italy
Lab web page:
Personal page:
We had the same problem and inevitably switched the bulbs to ones that wouldn't create that problem. Don't know if that's a possibility for you.


Dear Damien,

Two potential methods to get around this problem are; one, use three-phase power to avoid all lights operating in phase hence causing the flickering; and two, use DC powered lights which don't have the phasing problem. DC lights can be powered by an AC power source and a transformer converts the AC into DC power.

Both solutions will be quite expensive as for method one it means getting major electrical work done in your lab to accommodate three phase power. For method two new lights would need to be bought with the corresponding electrical work to go with them.

While not cheap solutions these are what we have installed in our lab.

All the best,


John Baker
Senior Biomechanist
AIS Movement Science - Biomechanics
Australian Institute of Sport
Australian Sports Commission

Tel: +61 2 6214 1347
Fax: +61 2 6214 1593
Mob: +61 417 239 143
Hello Biomech-L subscribers

We ran a series of tests where we used high speed videos at 500 fps.
The ambient light could not be turned off, therefore our videos show a
huge variation in brightness from frame to frame. The ambient lights
flicker at 120 Hz.
I tried various filters available for virtualdub (such as deflick and
msu_deflick) to remove the flicker, but could not see any significant

This is not a new issue, but I could not find any posts in the
Biomech-l archive regarding methods/tools to remove flicker from
digital videos, after they were taken. If you have any suggestion or
experience, I'll be grateful if you could share them,