methods for differentiation between trees and grass
1 view (last 30 days)
Show older comments
Ajad Singh
on 10 Feb 2022
Commented: Image Analyst
on 19 Feb 2022
Im looking for methods to use to help differentiate roadside trees on a motorway from grass (as shown in the picture), which then later i can use this information to analyse the pixel values of the trees. (im not asking for any code, just ideas which i could maybe try and implement)

0 Comments
Accepted Answer
Image Analyst
on 10 Feb 2022
Why? Why do you care about the nature off the road? Like what kind of plant it is? Just curious.
Anyway you could try a combination of color thresholding and texture segmentation.
Alternatively you could try SegNet deep learning but you'd have to outline/label dozens or hundreds of photos manually.
4 Comments
Image Analyst
on 11 Feb 2022
If it's bolted down I don't think it should change so much that your manually defined regions would change a great deal. So after taking the mean color of all the pixels inside the mask regions you'd have a pretty good idea of how the mean color changed with the seasons.
Image Analyst
on 19 Feb 2022
SegNet demos attached. Adapt as needed.
More Answers (1)
Benjamin Thompson
on 10 Feb 2022
That is a very complicated problem. See the example Semantic Segmentation Using Deep Learning in the Computer Vision Toolbox for a similar problem.
3 Comments
Benjamin Thompson
on 10 Feb 2022
Try it and see. Remember you need a lot of data for this type of solution to work on full motion video. Trees and grass can look a lot different during different weather conditions, time of day, types of trees/grass species, whether the grass has been recently mowed, etc.
See Also
Categories
Find more on MATLAB Support Package for IP Cameras in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!