I want to extract the number of plants per row in a crop land. I have and RGB mosaic from UAV orthorectified aerial images.

I would like to be able to identify single plants in order to count each plant.

closed as too broad by PolyGeo Sep 26 '16 at 2:30

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • spatial resolution? plant specie? – Pau Feb 10 '15 at 11:51
  • Welcome to GIS SE! As a new user be sure to take the Tour to learn about our focussed Q&A format. I recommend not thinking about GIS SE as being some sort of online GIS tutor. For your questions to be answered here they should as much as possible describe not just what you want to do, but precisely what you have tried and where you are stuck trying that. – PolyGeo Sep 26 '16 at 2:30

You should do a Supervised classification. Where you have defined polygons you can convert them to a centroid. Depending on the number of plants that touch, you will process them differently. Take note, that from experience doing classifications from RGB imagery for vegetation produces poor results.

  • I second that regarding using RGB imagery for feature extraction. The OP's best bet will be to use OBIA methods in packages such as eCognition, SPRING, or Orfeo toolbox in QGIS. – Aaron Feb 7 '15 at 22:27

Would it be possible to acquire a near-infrared image? That would be very advantageous (if not even crucial) when extracting vegetation information, and acquiring such an image would be my first step.

Second, I would also recommend using a supervised calssification. You are potentially looking at very small patches that occupy a very narrow range in the electromagnetic spectrum, and an unsupervised classification might not even consider the patches (i.e. put them into another class). You also have not specified whether you are looking at the same plant or different kinds. The spectral signature could differ dramatically from plant to plant.

In the end, once the classification has completed, you can go ahead and vectorize the result. Once you have that vector you can query it and discard all of the classes that are not crops (or just count the latter right away).

There are numerous software you could use to execute a classification. From the tags, I assume you might be using ArcGIS. If so, then be aware that you need to have Spatial Analyst installed and checked. If you do not have access to that you might want to look into QGIS, which has some remote sensing tools, including classification. However, from talking to my colleagues, some of these tools can be buggy (i.e. you need to provide inputs in a very specific way sometimes, and if you do they still might not even return the same results). I have done classification in ERDAS IMAGINE before and it worked very well (a very popular software at universities; maybe you can go ask somewhere if you can use it for a little). Although I have not used them, both IDRISI and ENVI would be an option as well for your task.

Not the answer you're looking for? Browse other questions tagged or ask your own question.