0
$\begingroup$

I have series of images extracted from a eye tracker video for both eyes. is there any image processing algorithm/method/technique available to find the following from those images:

  1. Horizontal eye movement for both eyes.
  2. Vertical eye movement for both eyes.
  3. Eye blinking for both eyes.

Thanks in advance for the help.

$\endgroup$
  • $\begingroup$ This depends on which types of images you have. Do you have close up images, such as iris datasets, or long range shots such as security, surveillance etc. Please elaborate. $\endgroup$ – Tolga Birdal Apr 20 '15 at 11:57
  • $\begingroup$ I am tracking eye movements with an eye tracker which is at a distance of 1m from the eyes. So i can say that my images are close up images. $\endgroup$ – uetmathematics Apr 20 '15 at 13:58
  • $\begingroup$ Could you share an image? $\endgroup$ – Tolga Birdal Apr 20 '15 at 14:23
  • $\begingroup$ Can you update/edit your question with queries made by others? There are good many algorithms that can do the job. It would make it a great question if you add the details in the questions asked by the comments. $\endgroup$ – Dipan Mehta Jul 15 '15 at 4:17
0
$\begingroup$

You can see for algorithms in ieee sites there are tons of algorithm which might suit your requirement and

An Algorithm for Real-Time Eye-Movement Tracking with Free Head Movement

Robust algorithm for video based eye tracing. Are some examples.

$\endgroup$
  • $\begingroup$ There is a problem with "tons of algorithms": the "tons". $\endgroup$ – Yves Daoust Oct 17 '15 at 15:50
0
$\begingroup$

If you’d like to come up with a rather quick solution by yourself you could try the following:

-Measure the retinas radius -do some preprocessing. Especially a BW transformation and maybe some thresholding. -Use hough circle to find the retina with the measured radius

The center of the circle can be tracked. Differences from previous sample would be the change in x and y direction. No circle for n seconds —> blinking time.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.