Estimation de ligne de base de capteurs d'humectation : int{\'e}gration et minimum locaux {\`a} diff{\'e}rentes {\'e}chelles

TL;DR

This paper introduces a new method for estimating the baseline drift in dielectric wetness sensors used in agriculture, utilizing L1 minimization and local minimum selection across multiple scales, validated on simulated data.

Contribution

It presents a novel baseline estimation technique combining L1 optimization and multi-scale local minimum selection for dielectric sensor signals.

Findings

The proposed method effectively estimates baseline drift in simulated sensor data.

It outperforms existing estimators in accuracy and robustness.

The approach is adaptable to different observation scales.

Abstract

Dielectric wetness sensors are used in agriculture to detect the presence of water on foliage and to predict the risk of disease development. The measured electrical signal has a base level drift that skews the alerts. We propose a method for estimating this baseline using L1 and selecting local minimums at an observation scale. The performance of the estimator is evaluated on simulated data and compared to the literature estimators.