best practice for increasing image processing speed

[aderogba]

Are there best practises for efficient image processing so that the overall perfomance speed will be improved? I have a need to do near real time image processing (threshold, filtering, particle analysis/cleaning and measurements) at 10fps. So far I am not satisfied with my cycle time so I am wondering if there are documented ways to speed up performance.

[GregSands]

I presume you've already turned on the Multi-Core Options.  Can you use some form of pipelining or parallel processing with your images?  Perhaps you could use different queues between each image processing operation, with dedicated worker loops for each operation that Dequeue-Process-Enqueue.  Perhaps something like one of these partially implemented ideas:

[aderogba]

Thanks. Please how do I turn on the Multi-Core Options? I am using IMQAdx.

[toto69]

Hi,

Navigates to the "Vision and Motion" palette, then "Vision Utilities" and "Image Management" palette, you will find ther the "IMAQ multi-core options" vi.

You will be able to set or get the number of cores used by the IMAQ functions.

Regards

[aderogba]

Will this function work for IMAQdx? Thanks.

[toto69 · Accepted solution]

Hi,

IMAQdx is only the driver, it's not directly linked to image processing while IMAQ is the vision library. This function allows you to use multi-core on the IMAQ function, to decrease the processing time., ecause image processing is the most consuming task for your computer.

Regards