Hi
I have a question and I hope to get useful answer
I'm working on a project to detect eye blink rate , can I do this using Arduino ? I'm new to this and I need some information
Regards
Hi,
The arduino can count, but it depends on what you want to use to detect the blink.
The arduino needs a sensor connected to it that signals the blink.
What is the application?
Tom.... 
Well, you need something to hold a sensor stably near an eye, so let's start with the frames of a pair of glasses.
Now the eye lash moves vertically, so breaking a beam of light (IR, optical, UV) makes sense. You could mount a little light emitter at the outboard side, but mounting a sensor on the other side, up against a nose, doesn't seem practical. So perhaps you just mount a reflector there, and read the interruption of the round trip part on an appropriately aligned sensor at the outboard side, with the light source.
Good luck, it's an ergonomically challenging project.
I believe there is facial recognition software which can monitor eye blinks. I'm not sure, but I thought I saw the eye blink software used with a Raspberry Pi.
Thanks for the answers . the application is to detect user blink rate and some processing on the result.
Is it possible for example to have IR sensor or EOG sensor on eye glass that send data by Bluetooth or WiFi to arduino or
raspberry pi for processing?
Thanks again
Hi,
The IR sensor is a good idea, I know that a lot of heat is emitted by the eyeball, but not sure how little it drops to in a blink.
Check the response time of an IR sensor/transistor and see.
Tom......
TomGeorge:
The IR sensor is a good idea, I know that a lot of heat is emitted by the eyeball, but not sure how little it drops to in a blink.
Check the response time of an IR sensor/transistor and see.
Most IR used with electronics is near IR. This is not heat.
The only hot things one can detect with near IR sensors are things which are almost "red hot". As an object heats up it will emit radiation in the far IR range (detectable with thermal imagers) but things don't glow in the near IR until they're almost ready to glow in the visible spectrum.
Near IR is better thought of as a color our eyes can't see. One would use near IR as one would use red light.
Passive Infrared(PIR) sensors detect human body temperature but most IR sensors do not.
@sarmad84, would you be allowed to place small stickers in the people's eyelids? If you could place sticker which reflected IR well you may be able to detect these stickers with a IR sensitive camera. The camera from the WiiMote might be good enough to detect when these stickers become visible (during a blink).
You might not even need a camera type sensor. As Tom suggested, IR sensors might be enough to sense the presence of a strong IR reflector.
The blink detector is probably the difficult part. One might put white clown makeup on the top of the eyelid which would be more reflective than the open eye, and use some type of photo transistor detector to detect the change in reflected light. I think line following bots use photo detectors that might be of use.
zoomkat:
One might put white clown makeup on the top of the eyelid which would be more reflective than the open eye,
I like the white makeup idea better than my reflective sticker idea.
You'd want to test the makeup to be sure it reflects IR as well as visible light. Some objects which appear opaque to visible light can be transparent to IR light.
DuaneDegn:
I like the white makeup idea better than my reflective sticker idea.
Why? Isn't skin a different colour than an eyeball? Unless you are a mime?
aarg:
Why? Isn't skin a different colour than an eyeball? Unless you are a mime?
I don't know if the color difference is enough to register with normal IR detectors or not. It would make things easier if the difference were detectable without needing to add anything to the eyelid.
IR can be seen with lots of digital cameras. One could shine an IR LED at one's skin to see how well it reflects. I'm not sure I like the idea of shining an IR LED at one eyeball. You'd want to make sure the LED was weak enough not to cause damage since the IR light won't cause the eye to blink nor will it cause the iris to contract.
DuaneDegn:
I don't know if the color difference is enough to register with normal IR detectors or not. It would make things easier if the difference were detectable without needing to add anything to the eyelid.IR can be seen with lots of digital cameras. One could shine an IR LED at one's skin to see how well it reflects. I'm not sure I like the idea of shining an IR LED at one eyeball. You'd want to make sure the LED was weak enough not to cause damage since the IR light won't cause the eye to blink nor will it cause the iris to contract.
Why shine anything at it? Just read the reflected light intensity from environmental illumination.
aarg:
Why shine anything at it? Just read the reflected light intensity from environmental illumination.
If the sensor is sensitive enough and the illumination consistent enough that would be great.
I personally would be surprised if simple IR sensors would be able to detect the difference between an open eye and a closed eye with ambient light. I'm often surprised though.
DuaneDegn:
If the sensor is sensitive enough and the illumination consistent enough that would be great.I personally would be surprised if simple IR sensors would be able to detect the difference between an open eye and a closed eye with ambient light. I'm often surprised though.
You wouldn't need an IR sensor. It could (maybe should) be a visible light sensor. A somewhat focused sense area around the eye, coupled with some rate change detection, could probably do it. I'm betting that the average light intensity there would create a recognizable pulse that you could filter (in the time domain) and detect.
My words aren't very precise now because I'm tired.
How would you change this so that the user could drive with one on? Obviously the Visible Light and IR sensors might not work so efficiently.
I am considering attaching a string to the eyelid and detecting a blink as something where the string is tugged a certain distance. Would this work?