Article
Hyperspectral Classification of Cyperus esculentus
Clones and Morphologically Similar Weeds
Marlies Lauwers
1
, Benny De Cauwer
1
, David Nuyttens
2
, Simon R. Cool
2
and
Jan G. Pieters
1,
*
1
Department of Plants and Crops, Ghent University, 9000 Ghent, Belgium; marlies.lauwers@ugent.be (M.L.);
Benny.decauwer@ugent.be (B.D.C.)
2
Technology and Food Science Unit, Flanders Research Institute for Agriculture, Fisheries and Food (ILVO),
9820 Merelbeke, Belgium; david.nuyttens@ilvo.vlaanderen.be (D.N.); simon.cool@ilvo.vlaanderen.be (S.R.C.)
* Correspondence: Jan.Pieters@ugent.be; Tel.: +32-9-264-61-88
Received: 7 April 2020; Accepted: 26 April 2020; Published: 28 April 2020
Abstract:
Cyperus esculentus (yellow nutsedge) is one of the world’s worst weeds as it can cause
great damage to crops and crop production. To eradicate C. esculentus, early detection is key—a
challenging task as it is often confused with other Cyperaceae and displays wide genetic variability.
In this study, the objective was to classify C. esculentus clones and morphologically similar weeds.
Hyperspectral reflectance between 500 and 800 nm was tested as a measure to discriminate between
(I) C. esculentus and morphologically similar Cyperaceae weeds, and between (II) different clonal
populations of C. esculentus using three classification models: random forest (RF), regularized logistic
regression (RLR) and partial least squares–discriminant analysis (PLS–DA). RLR performed better
than RF and PLS–DA, and was able to adequately classify the samples. The possibility of creating an
affordable multispectral sensing tool, for precise in-field recognition of C. esculentus plants based on
fewer spectral bands, was tested. Results of this study were compared against simulated results from
a commercially available multispectral camera with four spectral bands. The model created with
customized bands performed almost equally well as the original PLS–DA or RLR model, and much
better than the model describing multispectral image data from a commercially available camera.
These results open up the opportunity to develop a dedicated robust tool for C. esculentus recognition
based on four spectral bands and an appropriate classification model.
Keywords:
reflectance; logistic regression; partial least squares–discriminant analysis; random forest;
yellow nutsedge; weed classification
1. Introduction
Cyperus esculentus L. (yellow nutsedge) is a perennial C4 weed of the Cyperaceae family that
originated from (sub) tropical areas and is listed as the sixteenth worst weed in the world [
1
]. In 1982,
C. esculentus was detected for the first time in Limburg, the easternmost province of Flanders (northern
part of Belgium) [
2
]. Since then, the species has moved in west through Flanders; it now covers
an estimated agricultural area of 16,000 ha and is still spreading [
3
]. Cyperus esculentus is also
spreading rapidly in Central Europe because of accidental introductions and subsequent expansion [
4
].
The species is hard to eradicate because of its enormous capacity for multiplying and spreading,
and its low sensitivity to control measures [
1
]. Cyperus esculentus produces seeds and hard tubers
at rhizome tips [
5
]. Tuber dispersal is generally regarded more important for the spread of this
species than seed dispersal [
6
]; a single mother tuber is able to produce more than 1900 shoots
and nearly 6900 tubers in an area of 3.2 m
2
in one year [
7
]. These tubers can stay dormant in
the soil for several years; laboratory analysis showed a half-life of 5.7 months for tubers buried at
Sensors 2020, 20, 2504; doi:10.3390/s20092504 www.mdpi.com/journal/sensors
