Image correlation & Edge detection on FPGA

[R.V.]

Hi,

I am using Basler camera & FlexRIO 7966R with 1483 Adapter module to acquire the images. I am able to acquire the images & performing centroid calculation on the acquired images.

I am using camera in 10 Tap mode, so I am getting 10 pixels in one clock cycle.

Now I want to implement some more image processing algorithm on FPGA (FlexRIO), these are Hole filling, Correlation & Edge detection (contour). All these functions are available in vision but not available in vision fpga (except edge detection), but the algorithm is not disclosed in the block diagram for these functions.

So I am not able to implement these function on FPGA, any help regarding for the algorithms of these function on LabVIEW FPGA will be appreciable

[James.M]

You can't fit a whole lot on FPGA (relative to Windows) and unfortunately a lot of Vision operations are incredibly complicated. You're dealing with thousands of pixels and relating them to each other which uses exponentially more logic. You can only flip and flop so much before you run out of flip-flops.

Hole filling and Edge detection are both process-intensive because each pixel, or groups of pixels, must be compared to its neighbor. Here's an overview of hole-filling algorithm functionality.

If you look through the list of FPGA supported Vision VIs, you will see they are really just all of the basic operations (Math, Filters, Color manipulation, etc.). Unfortunately, this is what you're left with unless you want to try your own algorithms out for hole filling or edge detection.