|
|
|
![]() |
|
|||||||
|
||||||||
|
|
Thread Tools | Rate Thread | Display Modes |
|
#3
|
||||
|
||||
|
Re: LABVIEW camera tracking
Yelk11,
You've picked a great tool to work with! After you use the "Simulate Aquisition" step to load some images into the software from a folder (you can't acquire live from the Axis camera, although many Firewire cameras will work), you will want to follow this general approach: Filter to get rid of unnecessary data: Use the RBG or HSL (or others) in the Color Threshold tool to filter out everything except the target. This works as a software filter, same as a physical filter. We could get just the target frame to exist in the image, and everything else is black. Using a "Lookup Table" with the "equalizer" will convert the result to an 8-bit B/W image. This gives you substantially less data to process Get geometry: Use the "Find pattern", "edge finding", or other tools to locate geometry of interest, whether its just finding the center of the target or measuring its dimensions. Use coordinate systems on those features along with the measurement tools to get the information you want. There are many approaches, and definitely research machine vision documents online, particularly on NI's website. They've developed some great stuff. The Vision Assistant is designed to help you develop an algorithm, and our programmers tell me that just about anything you can do in LabVIEW you can convert to code. I assume its magic. Code and I don't mix....I am by no means an expert, but we've gotten some great results using this general outline. |
| Thread Tools | |
| Display Modes | Rate This Thread |
|
|