Uncertainty Quantification and Visualization for Optimal Decision-Making in Hyperspectral Imaging-Based Precision Agriculture [Roadmap for Measurement and Applications]

Abstract

Precision agriculture is rapidly transforming the production workflow for crop monitoring and food quality control through the use of state-of-the-art instrumentation and measurement (I&M) methods that gather information about the crop rapidly and non-destructively. Specifically, the use of hyperspectral imaging (HSI) systems provides the capability to acquire not only spatial but also spectral details. It is no surprise therefore that HSI has become a notable instrument in many areas such as remote sensing and agriculture [1]. Yet despite the usefulness of the instrument, extracting relevant or useful information from HSI data is not an easy task, particularly in uncontrolled lighting conditions which affect the measurement of spectral responses of the object-under-test. In recent years, advancements in Machine Learning (ML) and Deep Learning (DL) have shown success in extracting useful features and performing complex pattern recognition across a range of problems. When used in combination with HSI, it has the potential to extract spectral responses related to plant phenotypes or chemical compounds, thus linking its spectral measurements to crop traits such as ripeness, onset of disease, and nutrient/water deprivation. Fig. 1 illustrates the four key computational steps involved in the use of HSI measurement systems: imaging, data processing, data analysis, and decision-making.

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.