I am using UAV data for mapping geomorphological processes in different environments, from coastal and estuarine subtropical areas to subpolar and polar glacial landscapes, and I want to profit from the huge amount of information in such high-resolution datasets. So, I was wondering if there are good free options for object-oriented image classification, alternative to eCognition for example?
Jorge,
Have you ever tried the OpenCV library?
https://docs.opencv.org/master/index.html
Best Wishes
Jorge,
I'd recommend an overview of Orfeo Toolbox or SPRING (INPE) for OBIA/Segmentation.
A few resources and further readings
https://www.mdpi.com/2072-4292/11/10/1238/htm http://gis.geog.queensu.ca/PublicationPDFs/2019_PHOTO_SegmentationReview_OBIA_final.pdf
and link to the tools
https://www.orfeo-toolbox.org/
http://wiki.awf.forst.uni-goettingen.de/wiki/index.php/Object-based_classification_(Tutorial)
http://www.dpi.inpe.br/spring/english/
All the best
Jorge,
Have you ever tried the OpenCV library?
https://docs.opencv.org/master/index.html
Best Wishes
Dear Paulo Eduardo Cardoso , Iago Costa and Dan Abudu
,Thank you very much for your inputs!
I appreciate that and will look carefully at your suggestions.
Anyway, I am still open to further recommendations, in case anyone else has other ideas...
Cheers!
Hi Jorge Arigony-Neto,
You could try SAGA GIS.
Good luck.
This is not quite object-based, but you might be interested in Buscombe and Ritchie's deep learning approach Article Landscape Classification with Deep Neural Networks
.The test datasets have focused on coastal environments, but we are testing them in alluvial and aeolian environments with some success. Buscombe has provided the python code to run this process along with some examples ( https://github.com/dbuscombe-usgs/dl_tools ). I'm not a programmer but I managed to get it to work...mostly. From experience, the system has a lot of potential, though I am hesitant to jump on the machine learning band wagon where I don't quite understand why areas are being classified. However, this does have the advantage of being able to deal with a lot of images.
For UAV data with topography, you might also be interested in our "process from form" approach Article Geomorphic Process from Topographic Form: Automating the Int...
We used arcpy, which is not open sourced but ESRI python code, however the concept could be applied to other GIS software. We designed it specifically for using imagery, repeat topography and topographic change patterns to inform about dominant geomorphic processes for use in a process specific sediment budgets. Our code and example data is provided as supplementary information to the paper and I have some code modifications for sub-meter data on my RG page.
Dear Jorge Arigony-Neto
As Paulo mentioned, Orfeo is god for what you have asked. But the eCognition still the best. You could try it. Additionally, ArcGIS Pro could do a wonderful job.
Good luck.
Jorge,
Take a look at GEOBIA (INPE site). It's is based on Terralib library (open source either) and dedicated to process high spatial resolution with Geographic Object-Based Image Analysis Algorithms
Dear Jorge,
Please, try ArcGIS Pro or SAGA SIG (https://sourceforge.net/projects/saga-gis/#)
Thanks
Hi Mansour take a look at Interimage and Geodma at INPE opensource softwares home page.
Hi mansur there is an extension to do this, designed by Hammond group for morphology, using DEM, You can ask to get it. and there is a tool script for this>
You could try with:
SAGA SIG :https://sourceforge.net/projects/saga-gis/
or Python:
https://www.mdpi.com/2072-4292/6/7/6111
Article A Python-Based Open Source System for Geographic Object-Base...
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping