Vous êtes sur la page 1sur 4

International Journal of Research in Computer and ISSN (Online) 2278- 5841

Communication Technology, Vol 5, Issue 2, February- 2016 ISSN (Print) 2320- 5156

Automatic Weed Detection System And Smart Herbicide


Sprayer Robot For Corn Fields
1
Aiswarya Mohan, 2Farza Parveen, 3Sneha S Kumar, 4Sneha Surendran, 5Tinu Anna Varughese, 6Nitha S
12345
UG students, Department of Electronics and communication Engineering ,
6
Assistant Professor, Department of Electronics and communication Engineering,
Sree Buddha College of Engineering for Women, Kerala, India

Abstract-The goal of this paper is to develop an sprayed only in some part of the weed in field The
automatic plantation system.Here fruit and weed in fruit and weed can be detected.The image processing
the plantation is detected and herbicide is sprayed for is the tool used for the identification.Color based
the weed affected area.The temperature of the field is segmentation procedure is carried out in the
monitored inorder to provide envelope to the field. MATLAB section.Twocolours are mainly used ,red
Inorder to achieve this goal the whole plantation is and yellow. The colour identified is red, then it is
divided in to two zones. A camera is used to capture assumed that a fruitred then it is assumed that a fruit
the top view of the field. Colour based segmentation is detected and a message is sent to the owner .If the
procedure for detecting the plants which are suffering colour detected is yellow, then it is a weed the pump
from protein deficiency. Here we consider the get activated and herbicide is sprayed to the field and
yellowing of plant leaves. So we process the video the send the owner the alert message. The
frames for segmenting the yellow coloured region. temperature monitoring of the field is by using a
When such regions are detected, we enable the temperature sensor. If the temperature is greater than
hormone pumps on corresponding zone by sending a threshold value then the covering of the field is
proper commands to the controller on the embedded opened[1].
control circuit. At the same time we can alert the
owner by sending SMS using GSM Module. We also II LITERATURE REVIEW
segment the image for RED colour which is
Automation in agriculture monitor various
considered as fruits on that region. In this case also
aspects of a field like fruit detection, weed detection
we inform the owner through SMS.
and temperature sensing of the field. This is a
Index term- Machine, Vision, weed detection, demonstration of how to achieve automation in
colour segmentation, MATLAB. agriculture with low cow cost and simple method. To
detect weeds in agriculture several method are
1 INTRODUCTION proposed. Feyaerts and van Gool [2] used a
spectrograph camera to identify weeds. They
Now a days, agriculture is a field which obtained up to 86% classification accuracy, but the
requires automation in its various applications. New scheme was not applicable to in-field purposes.
advanced technology is used in the agriculture Another approach for weed detection was proposed
operation due to labor shortage increase in labour by Nielsen, M., Andersen. This demonstration is
cost etc. Agriculture operation needs automation, mainly focus on tomato plants. Nielsen analyzed
among this one is weed control . In the conventional images acquired by a stereoscopic vision system, but
weed control system, herbicide is sprayed uniformly ground irregularities had negative impacts on
over the field which may damage crop condition. . results[6].There were some other classification based
Apart from that, there is negative impact on the plant, on height differences. For identifying broad-leaved
soil large amount of herbicides are wasted and it is

www.ijrcct.org Page 55
International Journal of Research in Computer and ISSN (Online) 2278- 5841
Communication Technology, Vol 5, Issue 2, February- 2016 ISSN (Print) 2320- 5156

patches in cereal crops Berge developed a method The system model consist of the following basic
based on shape parameters and as a result achieved systems:(i)fruit and weed detection,(ii) mat lab
84 to 90% classification accuracy. Sogaard also used section control unit, (iii) wireless transmission system
active shape models as a criterion to classify weed (iv) embedded section ,and (v) herbicide spraying
species. But the problem of using height as a criteria system.
is that it needs ideal conditions; shape of the leaf
should be well displayed [7]. Moreover, because of a.Matlab Section
being shape-based and also wide range of species,
The fruit and weed detection is the first step by using
they cannot be developed to be used for all of them.
a camera. First of all the whole field is divided in to
Another option for classification is based on textural
two zones. The image acquisition is the first step that
information of weeds and crops. Machine vision
takes place in MATLAB.The acquisition can be done
systems present a great potential to be used on data
by any type of digital camera such as normal
collection for precisionagriculture, where images
webcams. The webcam should be installed
would be used to extract information.
perpendicular to ground. The output images of
Promissory results on automated weed thewebcam are in RGB format with a size of 640*480
detection were obtained using an image processing pixels.After that the acquired images are processed in
method .Another approach for taking images of a MATLAB.Colour segmentation is done,here two
field by using CCD camera mounted on a tractor colours are detected red and yellow[3,4].
[5]. With the popularity of digital cameras, personal
computers, image processing software, and global
positioning systems, machine vision systems could be
an option to cost reduction in data collection for
precision agriculture. Georeferencing the data could
be possible to identify spatial variability on weeds
distribution and allow site-specific control , using
patch spraying where herbicides are applied only on
the areas where the weed population economically
justify the application. Another option is to vary the
herbicide to maximize the weed control. The weed
maps could also be used as an aid to define
management zones. This work is based on the
hypothesis that the use of image processing methods
can facilitate data collection and identification of
spatial variability of weeds [7].

III PROPOSED METHOD

Thismethod proposesasystem which iscapable of


detecting fruit and weed on the basis of colour
based segmentation. This system creates an ewera
in the automation system. It consists of ZigBee Fig 2. Embedded section
based control system that transmits the wireless
When the weeds aredetected, enable the hormone
signals according to the input being selected based
pumps on corresponding zone by sending proper
on colour detected by the MATLAB section.
commands to the controller on the embedded control
A.System Model circuit. At the same time we can alert the owner by
sending SMS using GSM Module. We also segment

www.ijrcct.org Page 56
International Journal of Research in Computer and ISSN (Online) 2278- 5841
Communication Technology, Vol 5, Issue 2, February- 2016 ISSN (Print) 2320- 5156

the image for RED colour which is considered as


fruits on that region. In this case also we inform the
owner through SMS.A temperature sensor is used for
temperature monitoring and to provide an envelope
for the plantation.

B. HARDWARE IMPLEMENTATION

The system can be divided in to three parts.Colour


detection part,transmission part and control part.The Fig 4ZigBeetransceivers
MATLAB section deals with colour detection and it
is transmitted in to the control part by ZigBee c. Control part
transmission.
In the control part here we use LPC 2148
a. Colour detection part microcontroller. LPC2148 controller is an ARM
based microcontroller. This is the final step of the
The video of the field is initially captured by using a proposed system. This section controls the system
camera. The video is then processed according to according to the input received from MATLAB
MATLAB codes. When red or yellow colour is section . The status of the system is monitored by
detected the snapshot of that image is taken . After using an LCD display.
identification of colours ,control signals will be sent
wirelessly to the control unit. The control characters
that send to control part are *a,*b,*c,*d for fruit
detection on first zone ,fruit detection on second
zone, weed detection on first zone,weed detection on
second zone respectively.

Fig 5 Control part

C . Software design
Fig 3Colour detection software
The software section includes colour
b. Transmission part
detection software and embedded system software
ZigBee transceivers constitute the design. Colour detection software design includes
transmission part. ZigBee transmitter is connected to capturing the video of the proposed system .When a
colour detecting part. ZigBee receiver is connected to colour is detected the snapshot of that image is taken.
the control unit. Each colour is having a particular threshold value.

www.ijrcct.org Page 57
International Journal of Research in Computer and ISSN (Online) 2278- 5841
Communication Technology, Vol 5, Issue 2, February- 2016 ISSN (Print) 2320- 5156

When a colour is detected it compares with that [3] Feyaerts, F., van Gool, L. (2001). Multi-spectral
threshold value and corresponding codes are sent to vision system for weed detection. Pattern
the control section. The colour detection is done with Recognition Letters, 22, 667-674.
the help of MATLAB software.
[4] Piron, A., Leemans, V., Kleynen, 0.,Lebeau, F.,
The embedded sectionsoftware is developed Destain, M. -F.(2008). Selection of the most efficient
by using Keil for ARM based LPC2148 controller. wavelength bands for discriminating weeds from
The devices connected to the microcontroller will crop. Computers and Electronics in Agriculture, 62,
provide corresponding actions based on the input 141-148.
received.
[5] Sanchez, A. 1, Marchant, 1 A. (2000). Fusing 3D
1V. RESULTS AND CONCLUSION
information for crop/weeds classification.
Automatic weed detection and smart herbicide Proceedings of the 15th International Conference on
sprayer robot based on ZigBee is a very useful Pattern Recognition (ICPR'OO) (Vol. 4,p.4295).Los
project in agricultural fields. Since its fully Alamitos, CA: IEEE Computer Society Press.
automated, time and money can be saved also less
number of labours can be implemented. The user will [6] Nielsen, M., Andersen, H. 1, Slaughter, D. c.,
be updated about field and crop. The weeds can be Granum, E. (2007).Ground truth evaluation of
detected and removed effectively.Additionally, a fruit computer vision based 3Dreconstruction of
detection technique is also there based on the colour synthesized and real plant images. Precision
detected. The colour detection is done using Agriculture, 8(1-2), 49-62.G.
MATLAB software. Temperature of the field is
controlled using temperature shield protecting the [7] Sogaard, H. T. (2005). Weed classification by
crop from temperature crisis. ZigBee is used to active shape models. Biosystems Engineering, 93(3),
transmit the data from the MATLAB controlled unit 271-281
to the embedded system. The wireless part of the
system has been implemented by using Zigbee RF
modules. Hence, the system is highly efficient and it
consumes low power. A motor is used to pump the
herbicide and another motor to control the
temperature shield. The presented system can be
rapidly implemented and provides personalized
services by wireless media.

REFERENCE

[1] Amir H. Kargar B, Ali M. Shirzadifar,“Automatic


Weed Detection System and Smart Herbicide Sprayer
Robot for com fields”, ISM International Conference
on Robotics and Mechatronics.

[2] A. H. KargarBideh, A. Shirzadifar, (2012) "Real-


time stereo-vision based herbicide spraying,
Proceeding of Iran's third International Conference
on Industrial Automation.

www.ijrcct.org Page 58

Vous aimerez peut-être aussi