Detect land cover and vegetation changes with NI Vision

[Nando88]

I have a task at hand that I think I would be able to solve with LabVIEW or any other programming language, if I just knew how to approach this problem.

What I currently have is a bunch of images from different years, of different types of landscapes. What I need to do is tell what are the differences between those images (if the land cover and vegetation has improved).

I want to know if there is a mathematical approach to do this, so that I am able to get the results I need.

Thanks for your help, and I really hope to hear from the experts here at the NI forum, so I can solve my task.

Maybe I could use color, in order to detect the differences, but I am not quite sure, please let me know your opinion.

Thanks.

[wiebe@CARYA]

Sounds feasible. Not sure if NI Vision is up to the task... OpenCV is.

The details depend for 99% on the exact given input and required output. We'd need a hole lot more information...

You'd probably end up using some sort of feature extraction to correlate the images, and maybe some sort of classification. Support vector machines, principle component analysis, stuff like that.

[Nando88]

Thanks for the fast response.

Just out of curiosity, why do you think that the ni vision module isn't up for the task?

I could use the OpenCV module from the vi package manager, since it is free, but I am not quite sure what would be the difference. I am leaning towards the vision module, since I have more experience with it, but if you say that OpenCV is better, then I should probably go with that.

Please let me know your thoughts on why would OpenCV be better than the vision module?

Thanks.

[Nando88]

I just read you previous answer again.

Thanks for the ideas.

[wiebe@CARYA · Accepted solution]

---

@Nando88 wrote:

I am leaning towards the vision module, since I have more experience with it, but if you say that OpenCV is better, then I should probably go with that.

Please let me know your thoughts on why would OpenCV be better than the vision module?

---

OpenCV has a bit more modern feature set. And it's a bit more difficult to handle in LabVIEW. I don't think the free OpenCV package has a lot to offer (haven't checked recently), but OpenCV itself has. I'm not really up to date about OpenCV in LabVIEW's current state. I think most people use it through Python.

I think for instance support vector machines, neural networks, principal components, and stuff like that isn't available in NI Vision, or at least not directly. On the other hand, I haven't found any OpenCV replacements for caliper tools.

I think OpenCV leans more towards AI, NI Vision towards process control and inspection. There's an overlap of course.