Teach me Vision

[Hitchhiker 42]

Also, remember to start with crude vision. Don't try to over complicate things before you get a basic form working. As to the actual details, I'm sorry but I can't help you. We work in LabVIEW with NI Vision Assistant.

[Andrew Schreiber]

I can walk you through the process I used to get started with OpenCV, or at least the steps I used.


Step 1: Compile a sample program - This may seem trivial but do it. It makes sure your environment is set up right.

Step 2: Load a static image and display it - Again, simple, but you'd be surprised at just how important small successes are in programming.

Step 3: Run a simple filter on your static image(I'm partial to threshold by a color) and display the result. - Not only does it get you an understanding of how to process images but you're going to do a lot of this anyway.

Step 4: Convert between different color spaces. Typical you have an RGB image, you should convert to HSV. Even if you don't need to it's good to know how to.

Step 5: Start writing your filter to find the desired feature. Typically this is a series of steps to isolate the feature. Ours was a simple threshold by and hue range this year.

Step 6: Extract Filter - For this I typically use the various contour functions.

Step 7: Filter your contours - maybe you got a light, maybe there's multiple targets. This isn't really vision just simple data cleaning.

Step 8: Interface with robot - ??? This is the hard part.

[TheModMaster8]

To add on to this question. How do you export the video feed into java run on eclipse? in other words what i want is to use grip for the image processing but do the calculation for a crosshair, shape detecting and alignment on the live video in eclipse and I don't want to use the smart dashboard or anything. if this is at all possible that is

[Hitchhiker 42]

Originally Posted by TheModMaster8

To add on to this question. How do you export the video feed into java run on eclipse? in other words what i want is to use grip for the image processing but do the calculation for a crosshair, shape detecting and alignment on the live video in eclipse and I don't want to use the smart dashboard or anything. if this is at all possible that is

I'm pretty sure that if you use grip, you need to send the data through network tables, which necessarily implies the use of a smartdashboard.

[TheModMaster8]

Originally Posted by Hitchhiker 42

I'm pretty sure that if you use grip, you need to send the data through network tables, which necessarily implies the use of a smartdashboard.

well darn. Can you get smartdashboard on a mac?