Hi, I'm new to the machine vision systems and I'd like to get your opinion regarding the execution time. I would need this inspection to run in less than 100 ms.
I have a monochrome ISC-1782 NI Linux Real Time Smart Camera running code made with Vision Builder 2015. The camera waits for a rising edge from a real time system, acquires 1920 x 1200 image then makes the following inspection steps.
-search horizontally white to black and black to white transitions (2 vertical lines).
-search vertically black to white and white to black transitions (2 horizontal lines).
-calculate the width of this newly found rectangle.
-search for non-white particles inside the rectangle.
-send test status (pass/fail) to real time system over ethernet.
The region of interest for the inspection is around 300 x 500 px. The real time system triggering the camera logs the time between trigger signal and camera response. It's currently around 140 ms. Does this sound normal? Is there any ways I could make it faster?
I haven't made the Vision Builder code myself but I needed to change some parameters with the evalution version. Could this have increased the execution time?
That execution time could be reasonable given what you've described. You can get a much better idea of where the time is spent by using the "Benchmark Inspection" tool under the Operate menu. Let it run for several iterations and you will get a report of the execution time for each step. You can then focus efforts on those steps that take significant amounts of time.