Light detection panel for laser - help

[peasantmagik]

I am designing a laser targeting system on which various control methods will be implemented and compared. The system consists of a laser pointer mounted on a two axis gimbal. The laser shines on a 30 cm x 30 cm white screen, behind which is a digital camera. The camera sends all information into Matlab, which computes the X and Y coordinates of the laser on the screen. There coordinates are then sent into dSpace, where the control is performed.

Problem: Right now the camera and image processing is working at about 13 frames per second, which is usually too slow to get accurate measurements of the position of the moving laser.

Anyone have any recommendations for relatively cheap solutions to this problem? The whole white screen / camera subsystem can be replaced if needed.

Here is a link to the project if anyone needs any further information. https://sites.google.com/a/temple.edu/nonlinear-control-of-laser-targeting-system/

 

[mdjensen22]

Where is your biggest hang-up in timing?
Is it due to the hardware, or is the software or communication protocol the bottleneck?
Have you tried isolating the various groups to see your reaction time?

 

[peasantmagik]

It is either the camera, the image processing, or both. We tried giving Matlab processing priority, but there is not much we can do with Windows running. Also, the camera is a pretty cheap webcam connected via USB. I assume upgrading the camera would speed things up, but is the camera really the best option? Maybe a solar / photodiode panel would work better? Or a camera with programmable image processing capabilities?

I know dSpace makes a camera that would do all of the image processing and connect straight to the dSpace board. This would completely eliminate Matlab. The problem is that the camera costs about $17000.

 

[mdjensen22]

Since you are using a webcam, do you have the the frame rate of it? It should have some published specs. In addition, you will also have a delay for the software to handle the communication and whatnot.

If timing is key, you could probably set up a panel of photodiodes or phototransistors with a micro to handle what you are looking for and it should be very fast in response. It might be a huge pain to wire up though considering you're coming from an off-the-shelf set-up.