Jump to content

Please give me some advice


Recommended Posts

My name is ttkx, from china.

how do you think i can identify the human face using IMAQ and if there is any labview example code or any other details which can help me please send me the necessarry information as early as possible.

I already got image from USB Camera with LabVIEW, I try some methods with Vision toolkit , but the result is unsatisfactory.

Link to comment
do you think i can identify the human face

I think it unlikely that you can be successful. It is hard enough to get these systems to identify bad parts on a PCB but the subtle aspects of a human, no chance any time soon, at least not with these toolsets. I have seen systems setup with some 20 years development and imaging or optics PhD's on staff where they kind of crudely work but there are many many pitfalls. Again, some work kind of crudely but are used at a few airports from what I hear.

Link to comment
I think it unlikely that you can be successful. It is hard enough to get these systems to identify bad parts on a PCB but the subtle aspects of a human, no chance any time soon, at least not with these toolsets. I have seen systems setup with some 20 years development and imaging or optics PhD's on staff where they kind of crudely work but there are many many pitfalls. Again, some work kind of crudely but are used at a few airports from what I hear.

thank you!

sorry,I don't clearly describe it. It is only the simple application. which person certainly doesn't need to examine, so long as examines the general position of person face and the eye condition, for example: Closes one's eyes or opens eyes. It is used for identifying driver weary condition .

I think it is possible to use the complexion similar skin color model for detaching face area. but I don't have method for getting skin color model. the skin color model include RGB ,then add offset color(RGB), according to that, detach face area.

Would you think my idea is feasible ?

Link to comment

Hello Ni Hao! :)

I don't know what you exactly want to identiy or just tracking the human face. If you just want to track human face with color, you can try out using the color spectrum if you want to use IMAQ vision, maybe you can first learn the object color (like human face color), that way you will get color spectrum for the human face, then you can search the whole image (maybe with the same ROI you used to learn the color object) to find the closest ROI where the match score is high. But this is not the best way. Normally people use Camshift method, like what OpenCV uses. I used to have a LabVIEW toolkit called ObjTracker that implements this Camshift method to track human face based on color, but we don't sell/support this toolkit anymore. Attached is a document about camshift method for your reference, hope it does not make more confusion.

thanks,

Irene

My name is ttkx, from china.

how do you think i can identify the human face using IMAQ and if there is any labview example code or any other details which can help me please send me the necessarry information as early as possible.

I already got image from USB Camera with LabVIEW, I try some methods with Vision toolkit , but the result is unsatisfactory.

Download File:post-42-1150925026.pdf

Link to comment
  • 2 weeks later...

If the application is basically open or closed eyes in a general field of view, I presume that the majority of weary drivers occurs at night time. Thus you will need also need lighting. A 1,000,000 candlepower spot would work nicely :blink:

Taking that route, why not make it simpler- find a wavelength that is not visible but eye safe, IR as a possibility. The chosen wavelength would need to also reflect nicely off/within the eye. This should improve your signal/noise ratio. Then it is basically down to finding the two bright reflecting areas and testing if they disappear (eyes closed). Don't even need a camera for that, photodiodes would suffice.

Too bad our eyes don't reflect as well as animals, ever shine a flashlight into a cat or dog or horses eyes at night? You can see them from many 100's of feet away. :thumbup:

Link to comment
Taking that route, why not make it simpler-

My father and uncle drove non-stop from Buffalo, NY to Miami FL in the late 1950s. To stay awake, the driver would place his fingers on the top of the window glass, then roll the window up to the top. If he started to doze off, his fingers would be pulled and wake him up. It worked back then, before they even had the PC!

Sometimes technology is not the answer...

Maybe instead of passively monitoring the state of the driver, you could create an application that would engage the driver. Failure to interact, a progressive slowing or change in response might indicate fatigue. Combine technology with psychology :question: I remember reading about an intelligence/puzzle plugin for the DS-Lite recently. Something like that?

Link to comment
My father and uncle drove non-stop from Buffalo, NY to Miami FL in the late 1950s. To stay awake, the driver would place his fingers on the top of the window glass, then roll the window up to the top. If he started to doze off, his fingers would be pulled and wake him up.

There were obviously no ladies in the car, otherwise that strategy would never have been accepted :D

Link to comment
Maybe instead of passively monitoring the state of the driver, you could create an application that would engage the driver.

In other words, a virtual child? :laugh:

"Are we there yet?"

"No!"

"Are we there yet?"

"NO!"

"Are we there yet?"

no response...

"WHAAAAAAAAH!!!"

That always keeps me awake... throw in the occasional, "I gotta go wee wee!" and you have yourself a good prototype.

Back to the nature of the post though, I wonder if the effect of red-eye could be used in dimly lit environments with a light source close to the lens of the camera? I'm not sure if red-eye effects are valid in the infra-red range, but it might help in detection since nothing inside the cabin except the human eye can reproduce this phenomenon, and it doesn't have the downfalls of human skin tone/facial form differences.

Just a thought...

Link to comment
  • 3 weeks later...
Taking that route, why not make it simpler- find a wavelength that is not visible but eye safe, IR as a possibility. The chosen wavelength would need to also reflect nicely off/within the eye. This should improve your signal/noise ratio. Then it is basically down to finding the two bright reflecting areas and testing if they disappear (eyes closed). Don't even need a camera for that, photodiodes would suffice.

Too bad our eyes don't reflect as well as animals, ever shine a flashlight into a cat or dog or horses eyes at night? You can see them from many 100's of feet away. :thumbup:

The only problem is how to deal with the driver glancing from side to side or any direction away from the detector... You'd get a lot of false alarms. The other problem is blinking. You'd have to sample long enough to avoid those..

JLV

Link to comment
The only problem is how to deal with the driver glancing from side to side or any direction away from the detector... You'd get a lot of false alarms. The other problem is blinking. You'd have to sample long enough to avoid those..

JLV

Head mounted detectors could be applied, camera would have a harder time maintaining the field of view too. Low pass filtering/hysteresis should assist in ignoring blinks :blink:

Link to comment
  • 1 month later...
  • 3 months later...
  • 2 weeks later...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.