Analysis and Application of Multispectral Image Processing Techniques

Applied to Soybean Crops from Drones Vision System

Evelio Gonz

alez

, Cristhian N

nez

, Jos

e Salinas

, Jorge Rodas

, Mariela Rodas

, Enrique Paiva

Yassine Kali

, Maarouf Saad

, Fernando Lesme

, Jose Lesme

, Luis Gonzalez

, Belen Maldonado

and Jos

e Rodr

ıguez-Pi

neiro

Unidad Pedag

ogica de Caacup

e, Universidad Cat

olica Nuestra Se

nora de la Asunci

on, Caacup

e, Paraguay

TECHA, Caacup

e, Paraguay

Laboratory of Power and Control Systems, Facultad de Ingenier

ıa, Universidad Nacional de Asunci

on, Luque, Paraguay

Instituto Paraguayo de Tecnolog

ıa Agraria, Caacup

e, Paraguay

EPCI Laboratory,

Ecole de Technologie Sup

erieure, Montreal, Canada

College of Electronics and Information Engineering, Tongji University, Shangh

ai, China

Keywords:

Drones, Multispectral Imaging, Digital Signal Processing, Precision Agriculture.

Abstract:

Drones are important in precision agriculture applications since they represent a new tool that can increase

crop production. In this context, the digital processing of the images obtained from multispectral cameras

integrated into the drones makes it possible to analyze the stress state of the crops, their vigor, a burned area,

among others. The latter are usually obtained through proprietary applications with very high subscription

costs. For this reason, this article presents the step-by-step implementation process of the different methods

or algorithms to be applied to multispectral images using the open-source Python programming language. We

use a soybean crop as an example of the application, and the results obtained from applying the digital image

processing algorithms are presented.

1 INTRODUCTION

Agriculture provides food and raw materials and

employment opportunities to a signiﬁcant amount

of the population. The agricultural potential of

any country contributes to the process of economic

development through different channels of inﬂuence

such as the growth of the farming sector as sustenance

for other sectors of the economy, contributions of

productive factors from the rural sector to other

industries, the assistance of the agricultural industry

to the reduction of poverty, the rural sector as a source

of the domestic market and the agro-export sector as

a source of foreign exchange (Cervantes-Godoy and

Dewbre, 2010).

Agriculture is of paramount importance for the

economic development of several countries and

guarantees the food security of its inhabitants.

However, there is still low penetration of new

technologies in agriculture in most developing

countries, which has repercussions in the lack of

efﬁciency, productivity, sustainability, optimization

of costs and resources (Puri et al., 2017). In

this context, the use of drones (also known as

unmanned aerial vehicles) as an ally of farmers has

recently appeared and consolidated. The use of

drone technology contributes to multiple processes

in agriculture because it will be possible to capture

important information and evaluate the conditions of

the monitored land and thus detect existing problems,

applying high deﬁnition camera technologies and

georeferenced information. for your exact location

(Stehr, 2015). The importance lies in the possibility of

detecting prematurely and efﬁciently diseases, pests,

and the possible effects that may occur in the future

about climate damage such as frost and droughts

(Patel, 2016).

Digital image processing using a drone’s camera

focuses on the possibility of having a versatile

photogrammetric solution, rapidly deployed in

remote locations and delivering detailed information.

It is possible to obtain an up-to-date photographic

coverage of the growing area of interest using a

high-resolution multispectral digital. The global

positioning systems can georeference the obtained

photographs (or video) and reconstruct a 2D or

3D model using computational tools (Jurado et al.,

2020). To perform the analysis of crops such as

González, E., Núñez, C., Salinas, J., Rodas, J., Rodas, M., Paiva, E., Kali, Y., Saad, M., Lesme, F., Lesme, J., Gonzalez, L., Maldonado, B. and Rodríguez-Piñeiro, J.

Analysis and Application of Multispectral Image Processing Techniques Applied to Soybean Crops from Drones Vision System.

DOI: 10.5220/0010615107070715

In Proceedings of the 18th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2021), pages 707-715

ISBN: 978-989-758-522-7

707

the Vegetation Index (VI), water stress (drought),

measure ﬂowering, make an inventory by zones or

plants, identify zones without production, etc. (Shah

et al., 2021).

On the market, there is commercial software

capable of processing multispectral images obtained

by drones. Among the best known and most famous

cases is Pix4Dﬁelds and DJI Terra. The ﬁrst one

offers the Green Normalized Difference Vegetation

Index (GNDVI), Leaf Chlorophyll Index (LCI),

Modiﬁed Chlorophyll Absorption in Reﬂectance

Index (MCARI), Normalized Difference Red-Edge

(NDRE), Normalized Difference Vegetation Index

(NDVI), and LSIPI2 indices. This software also

allows us to conﬁgure any other index as long as it

only requires the red (R), green (G), near-infrared

(NIR), and red-edge (RE) bands of the spectrum.

Another similar software is the DJI Terra, which

offers the NDVI, GNDVI, NDRE, LCI, and OSAVI

indexes. However, both the Pix4Dﬁelds and the DJI

Terra are paid applications under the subscription

model; For the ﬁrst case, the annual cost at the date

of publication of this article is USD 2,000.00, while

for the second case, the yearly cost of the basic plan

is USD 1,199.00.

We ﬁrst identify the most popular methods and

algorithms applied to multispectral images obtained

by drones and applied to crops. The main contribution

of this paper is applying step-by-step of the most

relevant algorithms in state of the art (e.g., NDVI or

GNDVI) using an open-source Python programming

language. The rest of the document is divided as

follows. Section 2 introduces the main essential

issues about drones. Section 3 presents the results

obtained by applying the different techniques using

multispectral images obtained from a soybean crop.

The Python codes applied to implement the various

methods are also presented in the same section. The

main conclusions and future work are shown in the

last section.

2 DRONES IN AGRICULTURE

2.1 Drones: Deﬁnition, Classiﬁcation,

and Applications

Drones are reusable aircraft that can maintain

autonomous ﬂight or pilot through the use of radio

control. When referring to the types of drones,

there are several sets of nomenclatures. They are

based on various parameters such as weight, designed

application, level of autonomy, type of operation,

whether civil or military or structural conﬁguration.

The latter encompasses several properties and

topologies that modify these aircraft (Nguyen et al.,

2020; Paiva et al., 2021). Multirotor drones

have several engines that generate thrust using the

propellers and can thus sustain in air. They are aircraft

that provide:

• Better maneuverability.

• Ease of use.

• Increased load capacity.

• Greater comfort for transport due to the

compactness.

On the other hand, the ﬂight ranges are lower and

present difﬁculties in recovering from engine failure

(Paiva et al., 2018).

The ﬁxed-wing drones use motors for propulsion

and stay thanks to the lift of the winds aloft. These

drones have a wide ﬂight range (especially in linear

ﬂights) and recovery controls in case of failures.

However, for landing and take-off, large areas are

not needed, there are difﬁculties in maneuvering,

and they are less compact. Finally, hybrid drones

are a combination of ﬁxed-wing and multirotor

conﬁgurations, thus inheriting the advantages of both

technologies (Segales et al., 2016). Still, the control

of these drones is quite complex (Kali. et al., 2019;

Paiva et al., 2019a; Gomez et al., 2020; Paiva et al.,

2019b; Kali. et al., 2018; Kali et al., 2018).

According to the classiﬁcations expressed above,

these aircraft are ideal for use in different areas of

agriculture. Some examples are the assistance to crop

pollination, automatic precision fumigation. The use

of multispectral imaging by cameras to measure the

analyzed crop is the main focus of this paper.

2.2 Multispectral Images Obtained by

Drones

Most of the drones available in the market typically

have mounted an RGB camera. These types of

cameras mount a sensor that measures the capacity

of light within the visible spectrum. That is, the

spectrum that the human eye is capable of seeing.

With an RGB camera, we will only capture and

interpret colors as we see them. Therefore, we can

only detect problems that are already visible to the

naked eye from an aerial view, such as areas with little

vegetation. There are other ranges of radiation in the

electromagnetic spectrum that go beyond RGB and

are of great importance for precision agriculture. To

see this type of radiation (the human eye is unable to

see them), we need a multispectral sensor (Mogili and

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

708

Deepak, 2018). Multispectral cameras have this type

of sensor capable of capturing various spectra of light.

They are small and can take values of up to 6 spectral

bands.

Figure 1: Examples of multispectral cameras.

From the multispectral images captured by this

type of sensor, different VIs can be calculated that

indicate the health and well-being of the vegetation.

These values represent some of the characteristics of

the plants that we analyze in detail in the next section.

3 IMAGING PROCESSING

TECHNIQUES

For this paper, we use a multispectral image of a

soybean crop in Rostock, Germany. Differents VIs is

analyzed for the beforementioned crops by applying

differents digital image processing techniques.

VIs are quantitative measurements based on

reﬂectance values that tend to measure biomass

or plant vigor. VI is a combination of different

arithmetic operations applied to other spectral bands,

used in a way designed to produce a simple value that

indicates the amount of vigor of vegetation within

a pixel. This makes it possible to estimate and

evaluate the health status of the foliage, based on the

measurement of radiation that plants absorb or reﬂect.

For instance, chlorophyll, which is the green

pigment in leaves, strongly absorbs visible light for

the photosynthesis process. On the other hand, the

cellular structure of the leaves strongly reﬂects the

light of the NIR band; the more leaves a plant has,

the more wavelengths of light are affected.

3.1 NDVI

NDVI represents an index that allows you to generate

an image showing greenness (the relative biomass).

This index takes advantage of the contrast between

the characteristics of two electromagnetic bands, the

absorption of chlorophyll pigment in the R band, and

the high reﬂectivity of plant leaves in the NIR band

(Drisya et al., 2018).

NDV I =

NIR −R

NIR + R

. (1)

This index is the most widely used because its

basic principle relies on that the spongy layers of

leaves reﬂect a lot of light in the NIR, in stark

contrast to most non-plant objects. When the plant

becomes dehydrated or stressed, the spongy layer

collapses, and the leaves reﬂect less infrared light

but the same amount in the visible range. The

mathematical combination of these two signals can

help differentiate non-plant plants and healthy plants

from diseased plants. (Esser A, ). The code with the

Python language to obtain NDVI is:

# We start with the import of the modules

# necessary

import rasterio

from rasterio import plot

import matplotlib.pyplot as plt

import numpy as np

import os

# {We read the R band}

imgPath = ’C:/Img/’

red = rasterio.open(imgPath+’IMG_0142_RED.TIF’)

# {We read the NIR band} %leemos la banda NIR

nir = rasterio.open(imgPath+’IMG_0142_NIR.TIF’)

# {We visualize the image}

plot.show(nir)

#{We convert to float}

red = red.read(1).astype(’float64’)

nir = nir.read(1).astype(’float64’)

# {Error handling in the division}

np.seterr(divide=’ignore’, invalid=’ignore’)

# {We calculate NDVI using numpy arrays}

# {Empty cells or cells}

# {no data is reported as zero.}

ndvi = np.where((nir + red) == 0., 0,

(nir - red) / (nir + red))

# {We plot the results with the colors}

#{Red, Yellow and Green}

plt.imshow(ndvi, cmap=’RdYlGn’)

# {We add color palette}

plt.colorbar()

3.2 GNDVI

GNDVI is an index used to estimate photosynthetic

activity estimates and is commonly used to determine

the water and nitrogen consumption of the vegetation

Analysis and Application of Multispectral Image Processing Techniques Applied to Soybean Crops from Drones Vision System

709

Figure 2: Results obtained when applying the NDVI

technique.

cover (Kross et al., 2015).

GNDV I =

NIR −G

NIR + G

. (2)

It is associated with the green color of the

vegetation (chlorophyll levels). It is one of the

most used to establish water and nitrogen absorption

levels in the foliage of the different crops. Nitrogen

is a primary structural component present in leaves

and is an indicator of plant health, which is why a

deﬁciency of this nutrient can cause low productivity.

GNDVI is resistant to atmospheric effects since it

has a more excellent range of wavelengths (from 560

nm to 840 nm) than the NDVI (from 650 nm to 840

nm). Moreover, GNDVI is ﬁve times more sensitive

to the concentration of chlorophyll-a. This VI avoids

the problem of NDVI saturation at relatively low

chlorophyll concentrations (it saturates, depending

on the species, with amounts as low as 2 µg/cm of

chlorophyll-a).

# Read band G

green = rasterio.open(imgPath+’IMG_0142_GRE.TIF’)

# We compute GNDVI using numpy arrays

gndvi = np.where((nir + green) == 0., 0,

(nir - green) / (nir + green))

3.3 SAVI

Soil Adjusted Vegetation Index (SAVI) is an index

that attempts to minimize the inﬂuences of soil

brightness by using a brightness correction factor.

SAVI is regularly using in arid regions where

vegetation cover is low.

SAVI =

NIR −R

NIR + R + L

+ (1 + L). (3)

Equation (3) also contains the parameter L (when

L = 0, SAVI = NDVI). L can take different values

Figure 3: Results obtained when applying the GNDVI

technique.

depending on the amount of vegetation in the area of

interest and vary between -1 and 1. A low value is

used in ﬁelds with a high vegetation density; a high

value should be used for ﬁelds with little vegetation.

# We coompute SAVI using numpy arrays

savi = ((nir - red) / (nir + red + 0.5)) * (1 + 0.5)

Figure 4: Results obtained when applying the SAVI

technique.

3.4 BAI

Burn Area Index (BAI) uses the reﬂectance values of

the R and NIR bands of the spectrum to identify the

areas of the ground affected by a ﬁre.

BAI =

(0.1 −R)

+ (0.06 −NIR)

. (4)

BAI constitutes a helpful parameter to

discriminate the area affected by the ﬁre over

other covers. SAVI uses the spectral ranges of

charred materials and ashes produced after a ﬁre,

recognizable by relating the reﬂectance values of the

R and NIR bands. Then, SAVI is quite helpful for

mapping burned areas, allowing clear discrimination

between burned and unburned, compared to other

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

710

spectral indices such as NDVI, SAVI, and GEMI.

SAVI can be helpful in countries with a dry season

or after the ﬁres that affect agricultural and forestry

crops.

# We compute BAI using numpy arrays

bai = 1 / (np.power((0.1 - red), 2)

+ np.power((0.06 - nir), 2))

Figure 5: Results obtained when applying the BAI

technique.

3.5 CIg

Chlorophyll Index - Green (CIg) is an index that

allows obtaining an estimate of the chlorophyll

content of the leaves from the reﬂectivity rate in the

NIR and G bands (Broge and Mortensen, 2002).

CIg =

NIR

−1. (5)

# We compute CIg using numpy arrays

CIg = (nir / green) - 1

Figure 6:

Indice de Cloroﬁla: CIg.

3.6 CIre

Chlorophyll Index - Red-Edge (CIre) is a vegetation

index that makes it possible to estimate the leaves’

chlorophyll content from the reﬂectivity rate in the

NIR and RE bands.

Clre =

NIR

−1. (6)

# We read RE band

re = rasterio.open(imgPath+’IMG_0142_RE.TIF’)

# We compute CIre using numpy arrays

CIre = ((nir / re) - 1)

Figure 7: Results obtained when applying the CIre

technique.

3.7 GEMI

The Global Environmental Monitoring Index (GEMI)

is a non-linear index for global environmental

monitoring from satellite images. It is similar to

NDVI but is less sensitive to atmospheric effects. It

is affected by the uncultivated land. Therefore, GEMI

is helpful for areas of sparse or moderate vegetation.

GEMI = eta (1 −0.25 eta) −

R −0.125

1 −R

, (7)

being eta:

eta =

2 (NIR

–R

) + 1.5 NIR + 0.5 R

NIR + R + 0.5

. (8)

nir2 = np.power(nir, 2)

red2 = np.power(red, 2)

# We compute theta

theta = (2 * (nir2 - red2) + 1.5 * nir + 0.5 * red)

/ (nir + red + 0.5)

# We compute GEMI using numpy arrays

gemi = theta * (1 - 0.25 * theta) - ((red - 0.125)

/ (1 - red))

Analysis and Application of Multispectral Image Processing Techniques Applied to Soybean Crops from Drones Vision System

711

Figure 8: Results obtained when applying the GEMI

technique.

3.8 MSAVI2

Modiﬁed Adjusted Soil Vegetation Index 2 (MSAVI2)

is a recursion of the MSAVI index and attempts to

minimize the effect of bare ground on the SAVI.

MSAV I2 =

[2 (NIR + 1)−

√

θ],

(9)

being θ deﬁned as follows:

θ = (2 NIR + 1)

−8 (NIR −R). (10)

# We compute MSAVI2 using numpy arrays

msavi2 = 0.5 * (2 * (nir + 1)

- np.sqrt(np.power((2 * nir + 1), 2)

- 8 * (nir - red)))

Figure 9: Results obtained when applying the MSAVI2

technique.

3.9 MTVI2

Modiﬁed Triangular Vegetation Index 2 (MTVI2)

is an index that allows detection of the leaves’

chlorophyll content at the tree canopy scale.

However, it is relatively insensitive to the index of the

area with foliage. It uses reﬂectance in the G, R, and

NIR bands.

MTV I2 = 1.5[1.2(NIR−G)−2.5(R−G)]

√

θ, (11)

where θ is:

θ = (2NIR +1)

−(6 NIR −5

√

R) −0.5. (12)

# We compute MTVI2 using numpy arrays

mtvi2 = (1.5 * (1.2 * (nir - green) - 2.5

* (red - green))

* np.sqrt(np.power((2 * nir + 1), 2)

- (6 * nir - 5 * np.sqrt(red)) - 0.5))

Figure 10: Results obtained when applying the MTVI2

technique.

3.10 NDRE

NDRE is an index used to estimate vegetation health

using the RE band. When evaluating the health of

crops in the middle and late stages of growth, it is

advantageous where the chlorophyll concentration is

relatively higher. Additionally, NDRE can be used to

map nitrogen variability in ﬁeld leaves to understand

the fertilizer needs of crops better.

NDRE =

NIR −RE

NIR + RE

. (13)

# We compute NDVIre using numpy arrays

ndre = np.where((nir + re) == 0, 0,

(nir - re) / (nir + re))

3.11 NDWI

The Normalized Difference Water Index (NDWI) is

an index that serves to deﬁne and monitor changes

in surface water content. It is calculated with the

NIR and G bands. The Normalized Difference Water

Index (NDWI) is an index that deﬁnes and monitors

changes in surface water content. Crops suffer from

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

712

Figure 11:

Indice de Vegetaci

on: NDRE.

severe water stress during times of drought or lack

of irrigation, so this index detects in time the areas

affected by lack of water and prevent negative impacts

on crops. It is also helpful for evaluating the risk

of ﬁre, determining moisture in the vegetation cover.

Higher NDWI values indicate sufﬁcient humidity,

while a low value suggests water stress.

NDW I =

G −NIR

G + NIR

. (14)

# We compute NDWI using numpy arrays

ndwi = np.where((green + nir) == 0, 0,

(green - nir) / (green + nir))

Figure 12: Results obtained when applying the NDWI

technique.

3.12 RTVICore

The Red-Edge Triangulated Vegetation Index - Core

Only (RTVICore) is used to estimate the area index

with foliage and biomass. RTVICore uses NIR, RE,

and G spectral bands.

RTV ICore = [100 (NIR −RE) −10 (NIR −G)].

(15)

# We compute RTVICore using numpy arrays

rtvicore = (100 * (nir - re) - 10 * (nir - green))

Figure 13: Results obtained when applying the RTVICore

technique.

3.13 SRre

Red-Edge Simple Ratio Index (SRre) is a vegetation

index used to estimate healthy and distressed

vegetation. The ratio of light is scattered in the NIR

and RE bands. Then, it reduces the effects of the

atmosphere and topography.

SRre =

NIR

. (16)

# We compute SRre using numpy arrays

srre = nir / re

Figure 14: Results obtained when applying the SRre

technique.

To summarize, Table 1 presents the output of

each technique according to the Calcination Index

(Cal-I), Geological Index (GI), and Chlorophyll Index

(Chl- I). On the other hand, Table 2 shows the

spectral bands used by each index. All the indexes

presented in this paper are the most used in practical

Analysis and Application of Multispectral Image Processing Techniques Applied to Soybean Crops from Drones Vision System

713

applications. However, the proper VI selection for a

speciﬁc crop is beyond the scope of this paper.

Table 1: Result of the algorithms.

Method VI Cal-I GI WI Chl-I

NDVI X x x x x

GNDVI X x x X x

SAVI X x x x x

BAI x X x x x

CIg X x x x X

CIre X x x x X

GEMI X x x x x

MSAVI2 X x x x x

MTVI2 X x x x X

NDRE X x x x x

NDWI x x x X x

RTVICore X x x x x

SRre X x x x x

Table 2: Spectral bands used by each index.

Method R G B RE NIR

NDVI X x x x X

GNDVI x X x x X

SAVI X x x x X

BAI X x x x X

CIg x X x x X

CIre x x x X X

GEMI X x x x X

MSAVI2 X x x x X

MTVI2 X X x x X

NDRE x x x X X

NDWI x X x x X

RTVICore x X x X X

SRre x x x X X

4 CONCLUSIONS

This paper has addressed the use of drones in

agriculture. Different algorithms used by commercial

applications have been analyzed, also incorporating

others. Then we have studied a soybean crop

and applied the different algorithms using an open

programming language. The programming process

for each algorithm has been presented to be applied

directly to readers and the results obtained have

shown their correct operation.

ACKNOWLEDGEMENTS

The authors developed this paper within the

framework of the research project called “Unmanned

aerial vehicles in applications for precision

agriculture for monitoring crops (PINV18-765)”,

co-ﬁnanced by the Paraguayan National Council of

Science and Technology (CONACYT) with FEEI

resources.

REFERENCES

Broge, N. and Mortensen, J. (2002). Deriving green crop

area index and canopy chlorophyll density of winter

wheat from spectral reﬂectance data. Remote Sensing

of Environment, 81(1):45–57.

Cervantes-Godoy, D. and Dewbre, J. (2010). Economic

importance of agriculture for poverty reduction.

Drisya, J., D, S. K., and Roshni, T. (2018). Chapter

27 - spatiotemporal variability of soil moisture and

drought estimation using a distributed hydrological

model. In Samui, P., Kim, D., and Ghosh, C., editors,

Integrating Disaster Science and Management, pages

451–460. Elsevier.

Esser A, Ortega R, S. O. Nuevas tecnolog

ıas para mejorar

la eﬁciencia productiva en vi

nas. 1(15):4–9.

Gomez, V., Gomez, N., Rodas, J., Paiva, E., Saad,

M., and Gregor, R. (2020). Pareto optimal PID

tuning for Px4-based unmanned aerial vehicles by

using a multi-objective particle swarm optimization

algorithm. Aerospace, 7(6).

Jurado, J. M., Ortega, L., Cubillas, J. J., and

Feito, F. (2020). Multispectral mapping on 3d

models and multi-temporal monitoring for individual

characterization of olive trees. Remote Sensing,

12(7):1106.

Kali, Y., Rodas, J., Gregor, R., Saad, M., and Benjelloun,

K. (2018). Attitude tracking of a tri-rotor UAV based

on robust sliding mode with time delay estimation. In

2018 International Conference on Unmanned Aircraft

Systems (ICUAS), pages 346–351.

Kali., Y., Rodas., J., Saad., M., Benjelloun., K., Ayala.,

M., and Gregor., R. (2018). Finite-time altitude

and attitude tracking of a tri-rotor UAV using

modiﬁed super-twisting second order sliding mode.

In Proceedings of the 15th International Conference

on Informatics in Control, Automation and Robotics

- Volume 1: ICINCO, pages 435–442. INSTICC,

SciTePress.

Kali., Y., Rodas., J., Saad., M., Gregor., R., Alqaisi., W., and

Benjelloun., K. (2019). Robust ﬁnite-time position

and attitude tracking of a quadrotor UAV using

super-twisting control algorithm with linear correction

terms. In Proceedings of the 16th International

Conference on Informatics in Control, Automation

and Robotics - Volume 2: ICINCO, pages 221–228.

INSTICC, SciTePress.

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

714

Kross, A., McNairn, H., Lapen, D., Sunohara, M.,

and Champagne, C. (2015). Assessment of

rapideye vegetation indices for estimation of leaf

area index and biomass in corn and soybean crops.

International Journal of Applied Earth Observation

and Geoinformation, 34:235–248.

Mogili, U. R. and Deepak, B. (2018). Review on application

of drone systems in precision agriculture. Procedia

computer science, 133:502–509.

Nguyen, H. T., Quyen, T. V., Nguyen, C. V., Le, A. M.,

Tran, H. T., and Nguyen, M. T. (2020). Control

algorithms for UAVs: a comprehensive survey. EAI

Endorsed Transactions on Industrial Networks and

Intelligent Systems, 7(23).

Paiva, E., Gomez-Redondo, M., Rodas, J., Kali, Y., Saad,

M., Gregor, R., and Fretes, H. (2019a). Cascade

ﬁrst and second order sliding mode controller of a

quadrotor UAV based on exponential reaching law

and modiﬁed super-twisting algorithm. In 2019

Workshop on Research, Education and Development

of Unmanned Aerial Systems (RED UAS), pages

100–105.

Paiva, E., Llano, M., Rodas, J., Gregor, R.,

Rodr

ıguez-Pi

neiro, J., and Gomez-Redondo, M.

(2018). Design and implementation of a vtol

ﬂight transition mechanism and development of a

mathematical model for a tilt rotor UAV. In 2018

IEEE International Conference on Automation/XXIII

Congress of the Chilean Association of Automatic

Control (ICA-ACCA), pages 1–6.

Paiva, E., Rodas, J., Kali, Y., Gregor, R., and Saad,

M. (2019b). Robust ﬂight control of a tri-rotor

UAV based on modiﬁed super-twisting algorithm. In

2019 International Conference on Unmanned Aircraft

Systems (ICUAS), pages 551–556.

Paiva, E., Rodas, J., Kali, Y., Lesme, F., Lesme,

J. L., and Rodr

ıguez-Pi

neiro (2021). A review of

UAVs topologies and control techniques. In 2021

IEEE International Conference on Automation/XXIV

Congress of the Chilean Association of Automatic

Control (ICA-ACCA), pages 1–6.

Patel, P. (2016). Agriculture drones are ﬁnally cleared for

takeoff [news]. IEEE Spectrum, 53(11):13–14.

Puri, V., Nayyar, A., and Raja, L. (2017). Agriculture

drones: A modern breakthrough in precision

agriculture. Journal of Statistics and Management

Systems, 20(4):507–518.

Segales, A., Gregor, R., Rodas, J., Gregor, D., and Toledo,

S. (2016). Implementation of a low cost UAV

for photogrammetry measurement applications. In

2016 International Conference on Unmanned Aircraft

Systems (ICUAS), pages 926–932.

Shah, A., Agarwal, R., and Baranidharan, B. (2021).

Crop yield prediction using remote sensing and

meteorological data. In 2021 International

Conference on Artiﬁcial Intelligence and Smart

Systems (ICAIS), pages 952–960.

Stehr, N. J. (2015). Drones: The newest technology for

precision agriculture. Natural Sciences Education,

44(1):89–91.

Analysis and Application of Multispectral Image Processing Techniques Applied to Soybean Crops from Drones Vision System

715