Semi-automated Analysis Of Aerial Citrus

Semi-automated analysis of citrus grove data from remotely-sensed devices can reduce groundsurveying for appraisal, crop recording, yield analysis, and grove management. Although theygenerally start with the same aerial infrared imagery for their data collection, grove managers andindependent appraisers each have their own independent requirements for data analysis. By making theanalysis of this imagery more efficient and robust, image understanding technology can providesignificant benefit to the entire citrus industry.The Phase I research effort submitted in this proposal will focus on the potential agricultural uses ofknowledge-based computer image understanding technologies that were originally developed formilitary, manufacturing, and medical domains. These techniques will extend the standard suite ofremote sensing tools to provide robust mechanisms for automatic identification and measurement ofindividual and/or hedge-row citrus trees from aerial color infrared image data. These new techniqueswill address difficult problems such as the isolation of individual tree crowns, automatic spatialregistration and normalization, delineation of groves, tree counting and classification. A follow-onPhase II program will expand the recognition criteria, analysis goals, tree species and remote sensingmedia in an integrated prototype system.