NI Vision Builder software setup

[shrinidhi]

I am trying to set up vision system to monitor flow of water and decane into a glass container. We have installed the NI Vision builder software and the camera is hooked to computer with fire wire cable. Decane is colored red and water is colorless. They both collect into the same container and separate out forming a decane-water interface. Increase in the height of interface and level of decane with time will be used to calculate the volume collected in the graduated cylinder. The camera is used to identify the interfaces and change in the interfaces height is used to calculate the volume of liquids collected.The container is 250 cc with 34 mm OD and height of 380mm. Both decane and water flow simultaneously into the collecting cylinder. The range of flow ratees vary from 0.333 cc/min to 2.0 cc/min.Can you help me  setup the vision data acquisition system for this purpose. i.e, the steps involved in setting up the software.

[Tejinder_G.]

Hi Shrinidhi,

You may need to decide which method is best suited for your application, but there are potentially a few different ways you could go about this.  Regardless of which method you choose, it will be important to setup a test setup that has proper backdrop/lighting etc. 

One way is to use the thresholding function to filter out particular pixel values.  For example, you could choose to keep all "red" pixels and count the number of pixels you have.

The other way is to use the edge detection function to find where the red liquid meets the clear water.  Once you find this edge, you can measure the distance from that point to the bottom of your cylinder.

I hope this helps!

[shrinidhi]

Tejinder,

Appreciate your reply. But which one would be effective an edge detection or thresholding?

As my rates are miniscule and the changes are very small? 

Thanks

Shrinidhi

Graduate Research Assistant

Louisiana State University

[Tejinder_G.]

I would probably go with edge detection because it will be easier to pick up an edge rather than treshold for pixels.  Also, I noticed that your graduated cylinder has white ticks all along the sides, which will mess up your thresholding

[shrinidhi]

Tejinder,

Thanks for your inputs. Will probably ask some more questions if I have any issue during my setup.

Thanks

Shrinidhi

[BruceAmmons]

My advice would be:

1. Put a white background behind the cylinder, or you will start picking up objects in the background.

2. See if you can find a cylinder with no markings on it, since the markings are going to block the view of the liquids.

3. Use an ROI that goes right up the middle of the liquid, and use edge detection to find the two edges of the red liquid.

This will probably give you the best measurements you can get with this setup.  There will still be issues due to perspective, but those are harder to avoid.

Bruce

[shrinidhi shetty]

Tejinder and Bruce,

This is what the setup looks like. My aim is to measure the volume of water and also the volume of decane(coded red), by measuring the height of each liquid. I use two options to calculate the the length of liquids by find straight edge and caliper. With find straight edge, I convert the acquired image to grayscale and then use calculate step to find both the liquid lengths. As a back up or to check the logged data, I use caliper. What I have been experiencing is sometimes the edge detection works and other time caliper works consistent, but have problems with making them run symaltenously. With one example being, eg.the decane colum with caliper ends up detecting only the left side of the edge and never recognises the right side of the decane edge. The vbai, developed by me is attached along with the one of the images. Please comment on the vbai.

Do you think what I am doing is good way to do it? What are the issues with the vbai so that I can make both find edge detection and caliper work symateneously.