Cameras

[Shiner 2]

After searching with minimal results I figure I will put it to you all. I am trying to find information on implementing a camera that will verify product selection based on color. Basically this is a mixed product and has several hues, the goal is to verify the proper colors are going into the right packages.

[Chris Elston 79]

Can you just use a color sensor? Like a Keyence color sensor instead of a color camera?

[Shiner 2]

looking at all options, already spoke to Keyence, meeting in person next week with a demo. The main concern is that the surface is porous, like a brick, and the rep was not confident that the sensor would accurately detect hue variance. Concern A: If an item has the same color mix, but is a darker hue, will it distinguish. Concern B: If external lighting variables change, the amount of recieved light could fool the sensor into thinking the hue is not correct. Right now the line is manually loaded, and this will probably not change. I was trying to avoid having, at the start of each run, a teach session where QC loads a sample in front of the sensor and that is recorded for process accuracy.

[Chris Elston 79]

I've used this: http://www.bannerengineering.com/en-US/products/5/Vision-Sensors-and-Vision-Lighting/14/Vision-Sensors/379/PresencePLUS-P4-COLOR-Series/ and this: http://www.cognex.com/support/downloads/File.aspx?d=1844&l=1033&m1=InSight&m2=3 I liked the 1100C mostly because I was partial to Cognex... Color cameras aren't so cheap. I hope you have a budget for it.

[Shiner 2]

Thanks, yeah it is a possibility that the cust will see the price tag and just continue to use manual labor. That was the concern with the color sensors, it the porosity causes inconsitency, they already have that in human error.

[tip2112 0]

I've done numerous color sensing applications in the past, (mainly for Automotive Interiors) and I offer this one piece of advice.....LIGHTING. If you cannot maintain a consistent light source (ie. block out ALL ambient sources,shadows, etc,..) then you WILL be doomed to fail, regardless of the sensor you use. What kind of input options do you have available? If you hope to do this with Discrete channels, then even the high-end sensors will probably not be up to the task reliably ,given the types of surfaces and variation you describe. There are options that I could recommend, but would need a little more information as far as what you have to work with from a PLC standpoint ,as well as anticipated budget. I have successfully verified automotive carpets (these have LOTS of variability in the Hue,Saturation and Intensity scales), for around $1000.

[Shiner 2]

There is no spec yet for a required processor type, nor has a budget been defined. The customer is looking at this project with year end increase in productivity in mind, and if we get close to the 10% increase we are shooting for, budget probably will not be too big a concern if it is reasonable. The material is a hardened concrete type product, with uneven and porous surfaces.

[Peter Nachtwey 15]

This will usually work when others fail http://www.deltamotion.com/products/color/ We ask for samples and test before the decision to buy.