OBIA4RTM

  An open-source tool for object-based image analysis for radiative transfer modeling using ProSAIL (Prospect5 + 4SAIL) free for non-commercial applications (research and education) under Creative Commons Attribution-NonCommercial 4.0 International Public License.
IMPORTANT OBIA4RTM is currently just a first prototype and will be continuously updated.

  OBIA4RTM aims for plant parameter retrieval – relevant in smart farming applications – by using radiative transfer models (RTM) and object-based image analysis (OBIA) that directly addresses actual user needs and policy demands in a highly efficient, flexible and scalable way. It uses optical satellite data (concurrently Sentinel-2) as input. The RTM approach makes the tool transferable and nearly globally applicable to a broad range of different crop types, while OBIA accounts for producing results on a per-object rather than per pixel-base. Image objects have the distinct advantage of being directly related to real-world entities such as single field parcels. Furthermore, results on a per-object base can be easily managed and shared via geospatial databases and web interfaces and refer also to the requirements of the Big-Data era. The basic idea of OBIA4RTM is to combine two widely used Remote Sensing analysis techniques: Biophysical parameter retrieval from optical imagery by means of radiative transfer modelling (RTM) and object-based image analysis (OBIA) concept. While RTM accounts for retrieving the most relevant plant parameters relevant in farming context (Leaf Area Index, Leaf Chlorophyll Content, etc.), the OBIA approach allows for semantic enrichment of spectral data by means of incorporating expert knowledge and advanced spatial analysis techniques. OBIA4RTM relies therefore on two main pillars: It describes plant spectra by means of physical equations that are universally applicable by using RTM and it introduces the concept of spatial autocorrelation to reduce redundancies and provide a more meaningful image objects by means of OBIA. It is thereby capable to provide vegetation parameter retrieval techniques that are not bound by temporal or geographic restrictions. Furthermore, OBIA4RTM directly addresses objects and, thus, human needs as humans tend not to think in artificial spatial units (i.e. pixels) but in terms of tangible entities such as single field patches or individual trees in an orchard.