First off, I know these questions come up fairly often. I did use the search, but my project comes with unusual restrictions, so please hear me out.
My goal:
I want to build a machine that follows movements of tagged objects (e.g. a head) with a camera.
Approach:
I want to continuously measure the distance between two points (3 dimensions; between 0,1 m and 2 m), accuracy of 1 cm. One of the points (A) is constantly moving, the other (B) is stationary. My Problem: The sending unit (A) should not be visible in the frame, so if I track for example a head, I would mount it on the back of the head, facing away from the camera.
What technology should I use? Most technologies I found (Ultrasonic, IR) seem to only work when there are no obstacles between A and B. But I am always going to have obstacles!
I think I need a method that is not directional and not affected by flesh and bones… Radio wave strength? RFID signal strength? A magnet?
Does the sensor (or sensors) need to be in the same location as the camera?
If not, place them at surveyed locations out of sight of the camera, perhaps behind the subject.
Use trigonometry to calculate the coordinates of the head relative to the camera.
Has anyone ever measured the time between sending and receiving an X-ray or Gamma ray signal? Also, that doesn't seem either easy to accomplish nor safe.
mikb55, what do you mean by "location"? What kind of sensors do you have in mind for this?
blar1234:
mikb55, what do you mean by "location"? What kind of sensors do you have in mind for this?
IR LED with diffuser strapped to the back of the person's head. Pulse the LED at 1kHz.
At three locations you have an IR photodiode (with amplifier ) mounted on motorised pan tilt gimbal controlled by an Arduino. Use synchronous detection, or use an analogue bandpass filter to extract the 1kHz signal.
Measure the xyz coordinates of the camera and photodiodes using a tape measure.
Program each Arduino to systematically scan the room. Report the angular measurements of the strongest signal. By triangulation calculate the xyz coordinates of the IR LED within the room, then calculate the distance from the camera to the LED.
Commercial systems just use video cameras and image processing on fast CPUs to obtain the angles, but the Arduino doesn't really have the memory or processing power to do this so you have to use mechanical aids. To get the resolution you'll need to use precision (expensive) pan tilt mechanisms.
blar1234:
Well, I was talking about mounting it to a person's head; let's say I implied it.
No, you said "a head" - no mention of a person. For all I know, it could be a head of broccoli.
The art of getting good answers lies in asking good questions.
AWOL:
No, you said "a head" - no mention of a person. For all I know, it could be a head of broccoli.
The art of getting good answers lies in asking good questions.
The sending unit (A) should not be visible in the frame, so if I track for example a head, I would mount it on the back of the head, facing away from the camera.
Do away with that restriction, and the problem is solved.
jremington:
Do away with that restriction, and the problem is solved.
I know. So how is that supposed to help me?
AWOL:
No, you said "a head" - no mention of a person. For all I know, it could be a head of broccoli.
The art of getting good answers lies in asking good questions.
Sorry then. Native English speakers seem to have an advantage there. Is there anything you would recommend now that you know it should not be harmful to people?
mikb55, thanks for the detailed answer! I will have to read into that a bit, it sounds quite complicated. But I guess that's what I signed up for
I saw one video of a guy measuring distance to a radio transmitter via wave strength. Would that be viable?
3 dimensions implies a reference frame. If it's the head that is at the center and detection (of what?) is done from behind then count the chance that you can do this as very close to none.
If you can make observations from unblocked external positions then it becomes good chance.
I had already come across that. It seems to be a combination of several technologies.
To clarify; I don't want to use the camera for tracking. I want to shoot images from point B for artistic reasons and I need to be able to hide the sending unit from the frame.
GoForSmoke:
3 dimensions implies a reference frame. If it's the head that is at the center and detection (of what?) is done from behind then count the chance that you can do this as very close to none.
If you can make observations from unblocked external positions then it becomes good chance.
So in other words, you're saying that it's not possible to track the movement if the view is obstructed?
blar1234:
So in other words, you're saying that it's not possible to track the movement if the view is obstructed?
It's about price and safety, xrays and gamma radiation are already off the list.
Possibly with a really strong magnet and no conductors between, a linear Hall sensor could detect it but at 2 meters the magnet would need to be very powerful. The same magnet would probably deviate an electronic compass.
Radio source can be pointed with a loop antenna except through some things. With 2+ antenna on a base line, you might get position.
Note that Pozyx requires at least 4 external elements (Anchors) to work in 3D.
Grumpy_Mike:
So why is that a problem? It does everything you want.
It's not a problem at all! I am just interested in the specific technology that 'goes through flesh and bones', because that seems to be the problem with my requirements.
@GoForSmoke, thanks for your detailed comments, you are really helping me out a lot. I don't need the position in 3D space, I just need the distance! Wouldn't one antenna be enough then?
blar1234:
It's not a problem at all! I am just interested in the specific technology that 'goes through flesh and bones', because that seems to be the problem with my requirements.
@GoForSmoke, thanks for your detailed comments, you are really helping me out a lot. I don't need the position in 3D space, I just need the distance! Wouldn't one antenna be enough then?
Maybe. It is something to try. It might not work at every distance, look up the work of Heinrich Hertz.
What goes through flesh and bone?
Magnetism, the power of which falls off at distance cubed (very fast) but can be detected even faintly.
Some kinds of waves, not just x-rays and gamma waves either but they might go around, not through.