On Thu, 2010-09-02 at 17:54 +0200, Matteo Sisti Sette wrote:

Hi,

I need to measure the rotation speed of a spoked wheel.

As an alternative to a suitable sensor + arduino, I was considering using a webcam placed very close to the wheel and counting the frequency at which the spokes pass in front of the webcam.

Has any of you tried anything similar? Do you have any suggestions?

My naïf idea is that if a (natural or artificial) source of light is placed in front of the webcam, when the spoke passes in front of it, it will reduce considerably the brightness of the whole frame or of a big portion of it.

I'm looking for something simple: if this eventually leads to a complicated artificial vision task, or if a good calibration is difficult to achieve, then I would rather go for a tachometer and an arduino...

I don't think that using a camera for this is the best approach. It certainly depends on a few parameters like expected rate and such, how suitable a certain approach is, but using a camera - even if the expected rate is lower than the frame rate of the cam - adds a huge computational overhead. For rates close and above the fps it is even more difficult if not impossible.

The arduino approach is likely easier to implement, depending on what kind of sensors you can use. If you use a light barrier consisting of an LED and a light dependent resistor (LDR), you just need an additional resistor for building a voltage divider that can be directly hooked up to an analog input of the arduino.

You could do it even without using an arduino at all: hook up the resistor/LDR to your sound card input and use [dfreq~] from zexy to measure the rate. I think you can't get a more precise setup than this (by using the high sample rate of a common sound card). I once used that to measure the speed of a turntable.

Roman