All the CMUCam tracking I'm aware of uses color blob detection. I'd think this would be a problem in a garage.
Maybe Chris doesn't have anything green in his garage so whenever he want to work he puts on a green t-shirt?
Can the CMUcam track with other means than by color?
No plan of attack as of yet; just my first thought. CMUCam can do more than blob detection though, but that is where coding prowess enters in.
I can't think of a sensor other than a camera that will track your movement (well, other than a scanning PING-type arrangement).
There are several TTL cameras and Arduino variants out there, it just seems logical (to me) that a camera would be the solution over PIR motion sensors.
There are several TTL cameras and Arduino variants out there, it just seems logical (to me) that a camera would be the solution. . .
I'm not aware of a camera plus microcontroller project that tracks an object without using either some sort of target on the object or with the object being a specific color. Do you know of one?
I do think this would be possible. I think it would require comparing an earlier image to a more recent image and detect which area of the image has had the most changes. This seems like a daunting task for most microcontrollers. I'd like to give it a try sometime but I'm not sure if I have the prowess to do so.
Parallax sells a laser range finder (LRF) which also has a camera. It seems it's a likely candidate to test ones prowess on blobless tracking. The LRF has eight unused Propeller pins broken out to pads on the PCB. I've soldered some headers to these pads planning to use the pins as an 8-bit data bus. I have a crazy idea of trying to store images from the camera in external SRAM. This is usually at this point in my planning where I realize I'd probably be better off using a PC to so this.
Anyway, I think trying to use a camera to track an moving object without some sort of target or color is an interesting idea but I don't think it's a simple one. I also don't know of it being done with a microcontroller.
Comments
No plan of attack as of yet; just my first thought. CMUCam can do more than blob detection though, but that is where coding prowess enters in.
I can't think of a sensor other than a camera that will track your movement (well, other than a scanning PING-type arrangement).
There are several TTL cameras and Arduino variants out there, it just seems logical (to me) that a camera would be the solution over PIR motion sensors.
So your recommendation is a CMUcam and coding prowess?
I think the thermal imager mention earlier or the Wii camera with some sort of target would also work.
I'm not aware of a camera plus microcontroller project that tracks an object without using either some sort of target on the object or with the object being a specific color. Do you know of one?
I do think this would be possible. I think it would require comparing an earlier image to a more recent image and detect which area of the image has had the most changes. This seems like a daunting task for most microcontrollers. I'd like to give it a try sometime but I'm not sure if I have the prowess to do so.
Parallax sells a laser range finder (LRF) which also has a camera. It seems it's a likely candidate to test ones prowess on blobless tracking. The LRF has eight unused Propeller pins broken out to pads on the PCB. I've soldered some headers to these pads planning to use the pins as an 8-bit data bus. I have a crazy idea of trying to store images from the camera in external SRAM. This is usually at this point in my planning where I realize I'd probably be better off using a PC to so this.
Anyway, I think trying to use a camera to track an moving object without some sort of target or color is an interesting idea but I don't think it's a simple one. I also don't know of it being done with a microcontroller.