Abstract

Remote sensing could become an important and economically feasible tool for agriculture only if specific modules are developed to differentiate, isolate and process each parcel. A landscape scene is normally composed of many parcels with diverse cropping systems. Each cropping system (orchards, herbaceous) normally requires specific image processing software in order to analyze the targeted biotic factors, such as weeds or pathogens, or abiotic factors, such as nitrogen or water content, of interest. Furthermore, the characterization of such factors often needs managing multi temporal series of images. So that, the full use of remote images at agricultural landscape level requires: 1) to isolate each agriculture parcel; 2) to process it accordingly with the growing crop and the targeted factor; 3) managing both process for a multitemporal image series taken at the same area; 4) to automate all the previous mentioned steps, parcel by parcel and factor by factor; and 5) managing the outcome information to obtain a geo-referenced site-specific input prescription map.

    The aim of this paper is to make general considerations to support the need to automate the remote images processing and to outline the contribution of our research group in the development of ENVI® add-on for this purpose, as follows: 1) AUGEO-2.0®, to increase the image georeferencing accuracy; 2) CAIR, calibration of multitemporal image series based on vegetative pseudo-invariant features. 3) CROPCLASS®, to isolate and assess individual agriculture plots before precision processing; 4) CLUAS®, for clustering and assessment orchards units; 5) SARI®, for sectioning and assessment images of annual/ herbaceous crops. 6) AMI, special module for the automatic management of the previous referred software. These modules are available for research purposes upon request.