Need help in detecting a door as an object

[Gg$16sant]

I am doing a project in my university where i need my robot to detect an object which is a door from a live video through a webcam and then capture that image. I am using myRIO for it. Could someone please guide me on how i could complete this task.

[cbutcher]

I think here you have a significantly sized project and have given very little information.

My guess is you might find it easier to get help if you break your task into smaller sections, work on them for some time, then ask for help with specific difficulties.

For example, you might consider first attempting to capture the video, and separately comparing suitable image frames to check for a door.

These two tasks can be easily separated but should become more approachable once you try them alone.

With regards to the myRIO, I'm not sure what you want from the system. FPGA programming is a little more complicated than desktop/RT LabVIEW programming. Do you need FPGA programming? If so, that's another thing you can separate out.

Good luck!

[Gg$16sant]

Thanks a lot for the response.

I've been able to get the webcam to capture a video as of now. But I'm not quite sure how to detect the door in real-time.

How am I supposed to separately compare suitable image frames to check for a door?

And regarding FPGA programming, I won't be needing it.

[Gg$16sant]

I need to capture an image from a live video (through a webcam) whenever a door is detected.

Can someone guide me on how i could do this. would be of great help since I am new to this field in LabVIEW.

[ML927]

Duplicate ?

[Gg$16sant]

I didn't get you. Could you explain in more detail?

[cbutcher]

So if you have the image capture working, capture some images of doors, and some images of not doors, and use those to test your detection VI.

[Frankay]

If you are using IMAQdx you may want to do a grab and pattern matching to find the door. The Vision Assistant ExpressVI could also be a good way to get started if you are new to LabVIEW Vision programming.

IMAQdx does require a separate license than LabVIEW, and can be downloaded with Vision Acquisition Software. You just need the version that is compatible with your version of LabVIEW

---

Aaron F.

National Instruments

[yea_likethecity]

Hello,

The other suggestions in this thread are great. Using Vision Assistant especially would be a really good starting point, since it makes Vision processing much more intuitive. As for specific techniques, I think pattern matching could be helpful to you. Pattern matching will allow you to define a reference image and detect similar images based on a similarity score. We have a very in-depth example online which demonstrates the technique:

Pattern Matching Example

We also have shipping examples which demonstrate how to use pattern matching. You can find them in the NI example finder. You can open the NI example finder by opening any LabVIEW window and clicking Help -> Find Examples...

When the example finder opens, it will be organized by task. You can find Vision processing examples by navigating the task tree to Toolkits and Modules -> Vision. You can find a folder here called Pattern Matching which contains several helpful examples, or you can explore our other Vision functions to see which one best matches your application. I hope this is helpful and I wish you luck on your project!