The article says they are able to identify and track the mosqutio with image recognition software, which is quite understandable. But how are they able to direct a low-powered laser directly at the mosquito after identification with the camera ?
What makes them able to know the coordinates to point the laser, just from the camera information ? Or is there something else I am missing ?
Thank you for any help !
I am trying to do something similar, thats why I am asking.
Check out the kinect. If you are attempting to track something with your computer and locate the exact position in a 3d realm I think you will need more than a camera and/or an arduino. I hope you are not attempting to track something so small.
This type of project is not likely to be handled by an arduino outside controlling the physical side. You have tons to work out before you are ready to attempt the physical side of this type of a project. Before you can aim at something - you have to know it's exact position. You can tell that without ever controlling an external device.
My advice - start with the PC .. the arduino won't do much for you here - at least not at this point. If you do find out want to control something - then figure out the desired specs of how fast it needs to move, how accurate and what range. Someone may be able to help understand if the arduino can handle the requirements once there is that level of detail available.
However I was more interested in finding out how they are doing it. I see that this might be the wrong forum to ask, but I thought someone here might know how they are able to point the laser on the mosquito.
Now that we had a way to see the mosquitoes,we could track them using image recognition andmachine vision software (OpenCV is an amazing software package that makes the process of track-ing moving objects trivial). Tracking allowed us tolock the bugs into the machine vision software’scrosshairs
So it uses a computer with a camara running the OpenCV package. They used galvanometers to direct the laser beam to the location in space that corresponds to the pixel they identified the mosquito in. This would have to be calibrated. A simple way of calibrating would be to use a white sheet or card then aim the laser at say three different places on the card. For each place they would compare the voltage going to the galvos with the actual pixel the spot was detected in to work out a volts per pixel for each X and Y directions. Once you have that you do the sums and get the galvo voltage for any pixel.
Funny article, scientific fantastic, or absolute technical nonsense (IMHO).
Answering the question, you can get coordinate (2D - XY) from image recognition software, it's just targeted image coordinates on the TV (monitor) screen. It's really easy.
Using two cam it 'd be possible to calculate Z as well, 3D system. Forget about mosquito 100 ft away, it require equipment the same complexity and price tag, as they use on spy satellites ( 2 mm mosquito looks the same as 2 m vehicle from 30 km distance).
Second part, how to deflect laser beam to specific position in space, there are a several ways, and mirror-galvanometer is the stupidest one for this task. Ultrasound wave in the crystals, I' think would works, but it require more accurate esteem of the resolution budget (time/precision/angle) etc.