Utomatic Imaging System Mounted On Boom Sprayer For Crop Scouting Using An Off

utomatic imaging system mounted on boom
sprayer for crop scouting using an off-the-

shelf RGB camera
Author links open overlay panelColton Campbell, Ahmad Al-Mallahi, William Watson
Show more
Add to Mendeley
Share
Cite
https://doi.org/10.1016/j.compag.2022.106690Get rights and content
Highlights
 •
Automatic imaging system on boom sprayer is developed using a consumer camera.
 •
The system, tolerant to spraying and vibration, captures HD images at 30 fps.
 •
It is powered by sprayer battery and operates according to the engine switch.
 •
The captured images clearly display foliar anomalies tractor speed of 12 km/h.
Abstract
This article describes an imaging system mounted on the boom of a self-propelled
sprayer, which was developed to automatically capture and store high-definition video
images, while handling agrochemical spray and machine vibration. This was
accomplished using an off-the-shelf camera (DSC-RX0M2, Sony) and two single board
computers. To enable automatic operation, the system was designed to start and
terminate along with the engine. All components were powered successfully except for
the camera which runs only on its own battery. This was overcome by delaying
powering certain hardware and software components of the system for different periods
to allow consistent initialization of the entire system. Accordingly, the current version of
the system starts imaging 15 min after engine start-up and guarantees performance for
3 h of consecutive video capturing. After several field trials, the imaging system became
able to save images with the required quality. The developed software is able to handle
images at a rate of 30 frames per second without delay. In order to handle dynamic light
conditions, the camera shooting mode was set to manual at a very high shutter speed.
This provided detailed images at both sunny and cloudy conditions. While the current
captured images proved to be adequate to start machine vision system development,
there is a setback when trying to get images at early morning or evening as they tend to
be too dark. This will be handled while upgrading the system towards a machine vision
for the purpose of automatic scouting.
Graphical abstract
1. Download: Download high-res image (130KB)

2. Download: Download full-size image
Introduction
Crop scouting refers to the process in agriculture of assessing the crop status, and
presence of various harmful infestations. Namely, this means evaluating the presence
of pest, insects, fungi, etc (Fishel et al., 2001). Traditional scouting methods entail
farmers physically monitoring the field. This fashion of scouting requires significant
investments in time and labour, and produces incomprehensive results (Théau et al.,
2020). Accordingly, machine vision designs with machine learning algorithms are
emerging to replace traditional scouting procedures.
Imaging systems, in industrial fields, have been developed for incorporation within
larger machine vision designs, in which cameras are used to capture images which are
processed by computer algorithms to obtain information and perform specific actions.
Beach et al. (2019) state that the ability to automatically identify and inspect objects is
important for controlling manufacturing processes. For instance, Wang and Ding (2010)
invented a methodology to detect the quality of a metal cap based on machine vision.
Also, Hermary et al. (2018) researched creating 3D models of lumber, adjusting for
parallax error using two perspective cameras. Yu et al. (2009) developed a machine
vision-based instrument for detecting missing instant noodle seasoning packets,
comprising a computer, at least one video camera, an image acquisition card, a light
source, a position sensor and a controller. In outdoor applications, imaging systems are
commonly developed in conjunction with automobiles. One example of this technology
are machine vision systems which identify traffic signs and prompt predetermined
mechanical responses (Schofield and Lynam, 2009). Similar technology is being used
to adjust the headlamps of vehicles according to the presence/absence of other
vehicles approaching in the opposite lane. (Bos and Schofield, 2002). A popular
application of imaging systems, in modern passenger cars, are reverse backup cameras
(McMahon and Bingle, 2004) accompanied by software which detects, among other
things, road edges. Cameras used in these systems are built-in to support the specific
applications, and usually designed after extensive testing on general-purpose off-the-
shelf cameras.
However, in early stages of research and development of new systems, off-the-shelf

digital cameras are typically used to establish imaging and machine vision systems. In
particular, commercial RGB cameras have been found a reasonable option as sensors
for machine vision. Manns (2019) used an RGB camera (Hero4, GoPro) as a substitute
for industrial slit lamp camera systems in ocular photography. Using inexpensive and
readily available parts, Manns found that “The GoPro slit lamp camera system takes
adequate images for gross findings but falls short of more expensive imaging systems
for more subtle images.” This was mostly due to the auto-zoom feature of GoPro, which
would often focus on the Iris. Despite this feature, the GoPro was able to capture higher
resolution video than its industrial alternatives. Manns notes that the GoPro system has
a 1.8 times higher resolution than the IDOC system, and 3.4 times higher resolution
than the Haag Streit system. This contrast is surprising considering the prices of the
GoPro system, IDOC system and Haag Streit system are $111.47, $11599.00 and
$12899.00 respectively.
In agriculture, RGB cameras have been used to collect per row images of early season
crops (Higgs et al., 2019). Higgs et al. explain that these images can be used to count
small Brassica seedlings under specific field conditions. Additionally, imaging systems
have been found to identity fruit for harvesting. A low cost RGB-D camera/IR sensor
(Kinect model 1414, Xbox) was used to capture image data of color and shape features;
information which algorithms used to identify apples (Nguyen et al., 2016). The authors
continue by explaining that their system could identify 100% and 82% of fully and
partially visible apples, respectively. Similarly, a machine vision system was developed
for creating a 3D map of mature broccoli heads while harvesting using a RGB-D camera
(Kusumam et al., 2017). While these methods show the ability of using an off-the-shelf
cameras to detect objects, at the individual plant level, in the field, they are all used for
specific periods of the growing season, either at the beginning of planting or while
harvesting. This is because of the difficulty in finding a suitable mechanism able to
move the imaging systems in the field during the growing season.
In order to enable in-season crop monitoring, scanning using cameras mounted on

drones or using satellite images has been discussed and tested intensively recently.
Yang and Hoffmann (2015) used an off-the-shelf RGB camera (D90, Nikon), GPS
sensor/trigger, and camera mount to capture images under various speeds, elevations,
and ground cover conditions. They explain that their system “can easily be installed to
any agricultural aircraft for airborne remote sensing”. Similarly, UAVs using RGB
imagery (NEX-5N, Sony) have been used to identify and evaluate the disease severity
of potato late blight, in addition to monitoring the spread of this disease (Sugiura et al.,
2016). However, Théau et al. (2020) state that unmanned aircraft system (UAS) crop
scouting is challenged by several phenomena such as non-homogenous reflectance of
electromagnetic radiation. Also, aerial images are often limited in the resolution of data
they acquire (Higgs et al., 2019). Ehsani and Maja (2013) explain that though research
has shown aerial imaging can be useful in agricultural applications, most growers
abstain for concerns of price, low-resolution and inconvenience. In general, aerial
imaging systems tend to have two detractions: high cost, and low resolution, high
distortion photographs, limiting the available data for image processing and machine
learning.
The existing literature demonstrates (see Table 1) that machine vision is already being
utilized in various industries, for the identification and analysis of objects. This image
data is already being used as inputs for mechanical control systems. Additionally, off-
the-shelf RGB-D cameras are being utilized for machine vision applications within the
agricultural sector. This means these systems have demonstrated a resistance to the
weather, dust, and environmental conditions agriculture necessitates.
In this work, the goal is to develop an imaging system, based on an off-the-shelf

camera, that can be mounted on a self-propelled boom sprayer. The sprayer has been
chosen as the mounting point because it is regularly driven across the field throughout
the growing season. The imaging system shall meet these design requirements: high-
resolution images, tolerance to machine vibrations and agrochemical spraying, and to
be scalable to real-time image processing. As such, the system should operate
automatically after the engine of the sprayer starts, capture images on-the go, and resist
environmental conditions that may affect functionality or image quality including
variations of sunlight conditions.
Section snippets
Imaging system components and assembly
Fig. 1 shows a block diagram for the components of the imaging system. One single
board computer (Raspberry Pi 4B 4 GB RAM, Raspberry Pi) is used to control two
servo motors; a rotational motor (DXW90, MagiDeal) and one linear motor (PQ-12-r,
Actuonix). Another single board computer (Delta 420, LattePanda) (“Latte-Panda”)
connects to a camera (DSC-RX0M2, Sony) and is used for camera control and image
capturing. This computer is low-cost and runs on Windows 10, which is necessary to
run the
System power and operation

On August 20, the first evaluation of the system for video taken on August 17 and 19
took place. At this stage, the focus on the evaluation was on the hardware; that is, how
many times the vision system corresponded to the operation of the sprayer, and
whether Latte-Panda was able to save all the images. Fig. 5 shows the times, on
August 17, the driver started the sprayer and the response of the vision system. The
figure shows that the imaging system started few minutes after the engine
Conclusion
In this article, the development of an imaging system mounted on a self-propelled boom
sprayer that can operate automatically using an off-the-shelf camera was demonstrated.
After a series of field testing, the system was able to provide high resolution video
images and save them in independent hard disk. The video-capturing software was able
to capture the images from the frame grabber on-the-go and can be the basis of a real-
time image processing program for machine vision.
The mounting
CRediT authorship contribution statement

Colton Campbell: Data curation, Investigation, Methodology, Software, Validation,
Visualization, Writing – original draft. Ahmad Al-Mallahi: Conceptualization, Funding
acquisition, Methodology, Project administration, Resources, Supervision, Writing –
review & editing. William Watson: Data curation, Investigation, Methodology, Software,
Validation.
Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may
be considered as potential competing interests: Ahmad Al-Mallahi reports financial
support was provided by McCain Foods Limited.
Acknowledgement
This work is supported by the Natural Sciences and Engineering Research Council of
Canada (NSERC) under the Collaborative Research and Development Grant – Project
(CRDPJ 543912-19), McCain Foods Limited, and Potatoes New Brunswick (PNB). The
authors would like to especially thank McCain Foods Limited for facilitating accessing to
the farm in Grand Falls NB and mounting the vision systems for tests during the growing
season.

Utomatic Imaging System Mounted On Boom Sprayer For Crop Scouting Using An Off

Uploaded by

Copyright:

Available Formats

Utomatic Imaging System Mounted On Boom Sprayer For Crop Scouting Using An Off

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Utomatic Imaging System Mounted On Boom Sprayer For Crop Scouting Using An Off

Uploaded by

Copyright:

Available Formats

utomatic imaging system mounted on boom

sprayer for crop scouting using an off-the-

1. Download: Download high-res image (130KB)

However, in early stages of research and development of new systems, off-the-shelf

In order to enable in-season crop monitoring, scanning using cameras mounted on

In this work, the goal is to develop an imaging system, based on an off-the-shelf

System power and operation

CRediT authorship contribution statement

Declaration of Competing Interest

You might also like