Image analysis methodology based on geographic objects (GEOBIA) using RPAS (drone) with RGB sensor

Abstract

Context: The present work presents a method for mapping vegetation, through a process of classification by geographic regions, called GEOBIA (Geographic Object-Based Image Analysis) considered suitable for classifying very high resolution images (very high resolution - VHR). It is possible to perform the procedure with any equipment that has a good quality RGB sensor and allows the execution of flight plan applications. Method: The method was developed based on open source software (open source) to avoid costs with licenses, at all stages, from the capture of images, elaboration of cartographic products, processing of classification by regions and conclusion through area calculations. . Result: The study was applied in four areas of interest, all in the Greater Florianópolis-SC region, containing portions of the ecosystem of Pioneer Formations - Vegetation with Marine Influence, also called restinga areas, whose main target of the classification was the mapping of the areas invaded by Pinus sp. The method proved to be useful for image classification in general, and can be used in the management of other exotic plant species, or even in other environmental applications.