Help on project - Electronic Circuits Projects Diagrams Free

guen_max · 18th December 2006, 04:50 PM

As our project, we have to make a conventional web cam(fixed) into an automatic. Like it can follow the user, focusing him/her if he tends to move. So, I would like to ask what sensor can you suggest that we use.

We were thinking it can move on the x,y coordinate...coz the rotating type (up to 360 deg)would be much difficult right?And also what if there are 3 people in front of the webcam?How will it know which one to give importance to? One solution I was thinking would be to make a program which will give the order to the webcam which person to focus..do I make sense at all?

Your suggestions are would be really appreciated...Thank you.

audioguru · 18th December 2006, 06:43 PM

Are you going to make artificial eyes and an artificial brain? Good luck.

vortechs2k · 18th December 2006, 08:55 PM

You could use some sort of a sensor array which would detect whether an object was within say 3 ft. or so and then focus on the nearest object...as long as the object wasn't 3 inches away. Do you get what I'm saying?

I think it would be helpful to know if this was an undergraduate or graduate project. it seems for an undergrad project you wouldn't want to get too in depth, but for a graduate project you just may have to create artificial eyes and a brain.

dknguyen · 18th December 2006, 09:01 PM

I think you are thinking about the problem the wrong way. You are thinking about what the camera does rather than how it will do it. Don't make the camera identify one, or 3 people and follow them. That's really advanced machine vision stuff. Try thinking colour tracking. You could try motion tracking, but it's too much to explain on the forum so I won't- basically comparing two images taken in the same position spaced apart on a human time scale (0.5ms maybe, since that's the speed that people move at rather than us), and then compare what parts of the image have changed and find some way and somehow figure out which direction the changed part of the image has moved in, and center the camera on that. The camera movement will be jerky since it has to do the comparisons half a second apart to be able to see the movements that people make (assuming they are large movements), but then it doesnt depend on colour.

When you speak about the motion, what do you mean by XYZ coordinates? Is the camera moving linearily in space? Or rotating to follow the user? If it's just rotating it's just as easy to do 360 degrees as 180 degrees...just change the software limits of rotation and use a better motor.

Sounds hard...good luck! I'd go colour tracking. Give it a tracking priority order for red, green, and blue so that when someone tries it out it will *look* like it's picked a random person to follow when it's actually just following a colour.

Sceadwian · 19th December 2006, 02:46 AM

The human being of importance is almost exclusivly going to be the one TALKING... Use a mic local to each possible source and calculate it's received volume into the probabilty the user wearing the mic is talking activly. Use an ultrasonic tone generator at the mic's source to output an AM modulated wave that coresponds to the users mic's average amplituted over the last 5 seconds. At ultrasonic frequencie stereo mic's can pick up location pretty activly, as well as distance if the transmition power is of a known rate. Given human being are relativly speaking the same height (or you can calculate differentials) You could aim the camerea at any speaking source.

guen_max · 19th December 2006, 05:34 PM

dknguyen,

For now, I don't want to tackle the identification part where 3 people are involved. Let's say we only have one user...so in this case a motion tracking would likely be best right?(note: we can't attach anything to the user)And this is not really in depth as to we need artificial eyes & brains...It's more like we have to attach something to the web cam to make it move...I really can't decide either in linear or rotating...one of my mates said we should just use stepper motors since we have them already...and oh we have to use the pic 16f84 microcontroller.

And when you said motion tracking(can we use the ultrasonic sensor?), the comparison of images merely involves programming right?and it will send the instructions to the device to adjust how much the motor should move to focus the user...I hope I do make sense here...i'm just not good with words =(

thanks

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Title	Starter	Forum	Replies	Latest
FINAL YEAR PROJECT!	erique82	Electronic Projects Design/Ideas/Reviews	18	15th April 2007 11:09 AM
Problem with my Project: Motorised volume control	zenigame211	Electronic Projects Design/Ideas/Reviews	20	24th March 2006 02:43 PM
A Microcontroller based Analogue Waveforms Analyzer Project	km	Electronic Projects Design/Ideas/Reviews	76	30th June 2004 07:40 PM
Newbie requests input, project advice!	Rian_Colorado	Micro Controllers	8	22nd May 2004 01:53 PM
I need a project idea based on my strength or interest. :)	janetsmith2000@yahoo.com	Electronic Projects Design/Ideas/Reviews	10	11th November 2003 10:33 AM

18th December 2006, 04:50 PM	(permalink)
guen_max	Help on project As our project, we have to make a conventional web cam(fixed) into an automatic. Like it can follow the user, focusing him/her if he tends to move. So, I would like to ask what sensor can you suggest that we use. We were thinking it can move on the x,y coordinate...coz the rotating type (up to 360 deg)would be much difficult right?And also what if there are 3 people in front of the webcam?How will it know which one to give importance to? One solution I was thinking would be to make a program which will give the order to the webcam which person to focus..do I make sense at all? Your suggestions are would be really appreciated...Thank you.

18th December 2006, 06:43 PM	(permalink)
audioguru	Are you going to make artificial eyes and an artificial brain? Good luck. __________________ Uncle $crooge

18th December 2006, 08:55 PM	(permalink)
vortechs2k	You could use some sort of a sensor array which would detect whether an object was within say 3 ft. or so and then focus on the nearest object...as long as the object wasn't 3 inches away. Do you get what I'm saying? I think it would be helpful to know if this was an undergraduate or graduate project. it seems for an undergrad project you wouldn't want to get too in depth, but for a graduate project you just may have to create artificial eyes and a brain.

18th December 2006, 09:01 PM	(permalink)
dknguyen	I think you are thinking about the problem the wrong way. You are thinking about what the camera does rather than how it will do it. Don't make the camera identify one, or 3 people and follow them. That's really advanced machine vision stuff. Try thinking colour tracking. You could try motion tracking, but it's too much to explain on the forum so I won't- basically comparing two images taken in the same position spaced apart on a human time scale (0.5ms maybe, since that's the speed that people move at rather than us), and then compare what parts of the image have changed and find some way and somehow figure out which direction the changed part of the image has moved in, and center the camera on that. The camera movement will be jerky since it has to do the comparisons half a second apart to be able to see the movements that people make (assuming they are large movements), but then it doesnt depend on colour. When you speak about the motion, what do you mean by XYZ coordinates? Is the camera moving linearily in space? Or rotating to follow the user? If it's just rotating it's just as easy to do 360 degrees as 180 degrees...just change the software limits of rotation and use a better motor. Sounds hard...good luck! I'd go colour tracking. Give it a tracking priority order for red, green, and blue so that when someone tries it out it will look like it's picked a random person to follow when it's actually just following a colour. Last edited by dknguyen; 18th December 2006 at 09:09 PM.

19th December 2006, 02:46 AM	(permalink)
Sceadwian	The human being of importance is almost exclusivly going to be the one TALKING... Use a mic local to each possible source and calculate it's received volume into the probabilty the user wearing the mic is talking activly. Use an ultrasonic tone generator at the mic's source to output an AM modulated wave that coresponds to the users mic's average amplituted over the last 5 seconds. At ultrasonic frequencie stereo mic's can pick up location pretty activly, as well as distance if the transmition power is of a known rate. Given human being are relativly speaking the same height (or you can calculate differentials) You could aim the camerea at any speaking source. __________________ "Because I be what I be. I would tell you what you want to know if I could, mum, but I be a cat, and no cat anywhere ever gave anyone a straight answer, har har."

19th December 2006, 05:34 PM	(permalink)
guen_max	dknguyen, For now, I don't want to tackle the identification part where 3 people are involved. Let's say we only have one user...so in this case a motion tracking would likely be best right?(note: we can't attach anything to the user)And this is not really in depth as to we need artificial eyes & brains...It's more like we have to attach something to the web cam to make it move...I really can't decide either in linear or rotating...one of my mates said we should just use stepper motors since we have them already...and oh we have to use the pic 16f84 microcontroller. And when you said motion tracking(can we use the ultrasonic sensor?), the comparison of images merely involves programming right?and it will send the instructions to the device to adjust how much the motor should move to focus the user...I hope I do make sense here...i'm just not good with words =( thanks