This paper deals with the problem of classifying the risk of infestation by weeds in a field using geoestatistics techniques, image analysis and fuzzy classification models. The main attributes used to describe the infestation include seed density, seed density patch, weed cover and aggressivity to produce seeds in each region. Seed density reflects seed production per area unit; seed density patch reflects the influence of the neighbouring seeds in a clustering; weed cover indicates the extension of the emergent weed plant clusterings; and, finally, aggressivity describes the percentage of occupation of species with high weed seed production capacity. Data for seed density, weed cover and aggressivity for the different regions are obtained from mathematical models. In this paper, a fuzzy classification system using the attributes described is proposed to infer about the infestation risks of crop regions by weed plants. Simulation results of the proposed risk classification system are presented to illustrate its use in the site-specific herbicide application.
fuzzy logic; geoestatistics; populational dynamics; maps; image processing
