wu :: forums - Print Page


    
      
        wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> medium >> Sampling Rate of the Human Eye
        
(Message started by: william wu on Jan 8^th, 2005, 1:35am)

Title: Sampling Rate of the Human Eye
Post by william wu on Jan 8^th, 2005, 1:35am

A question I thought up recently and have been wondering about:

Does the human eye behave like a sampler? If so, how would you devise an experiment to determine its sampling rate?

Assuming that it does behave like a sampler, below I have two proposed experiments (feel free to think up an experiment of your own first before reading the proposals):

Experiment 1) Flash a white square on a black screen at gradually increasing frequencies until we reach some critical frequency f_c, at which the square no longer seems to be flashing to the viewer. Then aliasing must be occurring, since the viewer's sampling rate is not high enough to resolve the flashes. By the Nyquist-Shannon sampling theorem, the eye sampling rate is thus f_eye = 2 f_c.

Experiment 2) Take 2 equal intensity light sources flickering at the same rate. Separate their oscillations temporally with a relative time offset. Gradually reduce their relative time offset until you perceive the sources to be in total unison. This is the critical separation in time T_c that is distinguishable by the eye, and thus must be close to the sampling period of the eye. So f_eye = 1/T_c.

Questions: Are these experiments equivalent? Are either of them theoretically correct? Or is the human eye just not like a sampler at all?

Title: Re: Sampling Rate of the Human Eye
Post by TenaliRaman on Jan 8^th, 2005, 2:36am

Hmm,
I guess it must be a sampler otherwise i would see my monitor flickering or delay between frames in a movie. (Ofcourse refresh rate of the monitor is also an indicator).

Quote:

Are these experiments equivalent?

I think not. We are not accounting here for the interference effects in the second experiment which still could be picked up by our eyes if the two sources are separated by a good amount of distance.So unless they are kept pretty close, we cannot count them as equivalent.

Quote:

Are either of them theoretically correct?

The first experiment is more likely to give correct result than the second i believe.

-- AI

Title: Re: Sampling Rate of the Human Eye
Post by rmsgrey on Jan 8^th, 2005, 7:56am

In my experience, it is possible for me to percieve a flickering visual input as steady right down to something like 5Hz (I haven't actually measured it directly). At the other end, I can become aware of a flicker on this 60Hz monitor.

It's worth bearing in mind for your experiment that a lot of perception takes place in the mind rather than the eye, so acts of will or states of mind can significantly alter perception.

Title: Re: Sampling Rate of the Human Eye
Post by BNC on Jan 8^th, 2005, 12:57pm

The vision system (eye+brain) performs an integration on what you see (but I don't recall the time constant). That is the reason you don't see flickers on the TV and other frame-based equipment.

It's also worth remembring that the eye is not uniform. The center of it has higher resolution as compared with the peripheral vision, while the latter has more "fast action" sesitivity.

Title: Re: Sampling Rate of the Human Eye
Post by Grimbal on Jan 8^th, 2005, 5:24pm

I thing I learned doing photos (on a previous-generation camera), is that to give the same blurrness in movement as the eye would perceive, you have to set it to 1/60s. If you make a picture of a waterfall, if you use a shorter time, you will see drops like frozen in mid-air. If you use a longer time, it will look like cotton or more like fog flowing down.

Title: Re: Sampling Rate of the Human Eye
Post by towr on Jan 9^th, 2005, 7:51am

on 01/08/05 at 01:35:44, william wu wrote:

A question I thought up recently and have been wondering about:

Does the human eye behave like a sampler? If so, how would you devise an experiment to determine its sampling rate?

Each photoreceptor has it's own behaviour, so they don't necessarily sample the world synchronously. For example, it depends on whether they are exposed to light synchronously. And also on the intensity of the light, whether they tire quickly and how long their refraction period is (which, again, can differ per receptor and depends among other things on tiring).

And even if they eyes where working as a perfect homogeneous sampler, the brain still needs to interpret the signals. And if the brain can at all help it, it will fill in blank spots; so even at frequencies where the eyes might still distinguish high and low intensity lighting, the brain may decide it would look better when averaged.
Though, the opposite does occur very sporadically. Where people percieve only one image of the world every few seconds or so, and consequently can't well estimnate the speed of objects, which is rather a problem in traffic.

Of course this doesn't make testing 'sampling rate' useless. It has for instance been proposed as a good test for measuring how tired you are (and whether driving would still be responsible). Using a method similar to the first experiment you describe, but using a small handheld device (which you could potentially have with you when travelling).
It was a long time ago when I saw it, but I think a rate of 20 Hz is about where you should stop driving, and 40 is peak alertness (on average).