Multi and hyperspectral digital imaging based techniques for agricultural soil characterization

摘要

Soil characterization and monitoring in agriculture represent the primary key-factors influencing its productivity and the quality of the produced products. A correct and continuous knowledge of agricultural soil characteristics can help to optimize its use and its degree of exploitation both in absolute terms and with reference to specific cultivations. Soil characterization is conventionally performed adopting integrated physical-chemical analyses based on soil portion (samples), properly sampled, classified and then delivered to specialized laboratories. Such an approach obviously requires a chain of actions and it is time consuming. In this work it is examined the possibility offered by multi and hyperspectral digital imaging based spectrophotometric techniques in order to perform fast, reliable and low cost "in situ" analyses to identify and quantify specific soil attributes, of primary importance in agriculture, as: water, basic nutrients and organic matter content. The proposed hardware and software (HW&SW) integrated architecture have been specifically developed, and their response investigated, with the specific aim to contribute to study a set of "flexible", and very simple, procedures to apply in order to be utilized to operate, not only in agricultural soil characterization, but also in other fields as the environmental monitoring and polluted soils reclamation.