Line Detection and noise issues

[diego90]

Hi all,

I am doing a project to detect white lines painted across a grass field using Labview 11.0 and Vision and Motion plug in. I have tried capturing a video (AVI) and processed it by using extract single colour plane, threshold, remove small particles, etc to make the image easier to calibrate. But the problem is noise that I pick up in the AVI, images that contain dead grass, casts of shadows, and light intensity distort my images and I cannot output a binary image of just white lines. Can someone teach me how to detect these white lines with so much variable to account for? I have two attached VIs, the first one works great for the most part. I can output a binary image of white lines in a bright field. The second VI is where I am having problems with those shadows. Thank you very much! Your help is very much appreciated.

[Cameron-T · Accepted solution]

The attached files are not opening. Can you try to reattach them?

[diego90]

Hi Cameron,

Okay, so I just attached each VI separately. Does it work now?

Thanks

[diego90]

Do you also need the AVI files? I can't seem to attach them as well.

[Cameron-T]

It seems like your VI simply preparing the image for processing. I recommend using VI's from the Machine Vision palette to help find the white lines. As a quick way to test which VI's would work best, you could use Vision Assistant. It allows you to create a script and shows you the results as you go. You can the use the VI's that were most effective in recognizing the white lines. I would start with the Edge Detection VI's, but I would not be surprised if a pattern match turned out to be useful. It really just depends on what the application is. Hopefully this was helpful.

[diego90]

Hi Cameron,

Yeah, my project will incorporate live camera feed where I have to process it. I have always just processed it through the Vision and Motion palette, as you say. I have not really played around with Vision Assistant. I just  Is the Machine Vision palette within Vision Assistant?

[Cameron-T]

Vision Assistant combines most of the vision VI's into a friendly UI. It contains most of the Machine Vision VI's. I recomend using it so then you can either set a script that works using the Vision Assistant, or use it to find out what low level VI's would work best for you.

[diego90]

Hi,

So I have been taking a look at it and I am having trouble running Vision Assistant. I am not used to it and don't really know how it is suppose to work. I have been trying to find example block diagrams online and can't find similar examples. I see that I can process a single image, not continous images from a recording. It also shows that I can process images like I was doing before through Vision Acquisisiton. But how is it suppose to look in the block diagram? Once I am done with my processing within Vision Assistant can I control variables through a front panel? Help will be much appreciated.

- Diego

[Cameron-T]

I am not sure which variables you are referring to. Can you be more specific? The Vision Assistant is going to be an Express VI on your Block Diagram. You can set what the inputs and outputs of it are by selecting "Select Controls" on the bottom right of the Vision Assistant window.

In order to get configured in Vision Assistant, it is helpful to load a sample image to run a template from. You can then find the lines on the template image. If you develop a script with enough robustness, it should then be able to find it on all future images that the image processing script is ran on.

The Vision Assistant will take however many images you give it. So if you want to run a video, it can do that, just keep in mind that it is looking at one image at a time in the video feed.