Site Loader

Weed species decreases the growth of the crop and reduce farm yields.
Weed competes with crops for sunlight, space and nutrients. There
are a large number of various kinds of weeds in agriculture. Weeds have a great
impact on the development of agricultural production and agricultural economy.
Farmers  have their own manual way of
finding the weed in the farmland and removing them. The manual way  they follow have more labor work and is time
consuming. To reduce manual way of finding weeds which involves in more
labor work and time consuming the proposed work has automatic way of detecting
weed in a farmland. The main objective is to detect the weed using biomass
sensor(ph sensors, load cell sensors) and digital image processing. The biomass
sensing and image processing are combined to obtain an accurate result. When
the sensed values of the biomass sensor exceed the threshold values it
automatically enables the camera where the camera takes the picture of the
particular farmland and detects the weeds using color detection algorithm in
image processing. The weeds detected area will be intimated to farmers through
short messaging service where the SMS contains the direction where the weeds
are present and the picture of the detected weeds will be spotted and posted in
the website.                                                              I.     
Introduction Paddy
productivity in Tamil Nadu has always been the second highest in the country,
next only to Punjab and rice cultivation is one of the most important
economical sector in Tamilnadu . Hence, the need to develop
technological tools to improve  the
management and productivity of the sector. One of the       problems in rice fields is weed plants,
due to the fact that these plants compete for water and nutrients reducing up
to 70% of  the field yield if not control
is carried out. Generally, it can beconsidered that critical periods for competence in any cultivation
with weed are at initial of growing and development states, for rice as found
in this period is between 0 and 75 days. Therefore the identification and
removal of this plants are very important for rice growers. In this paper we focus on the weed identification problem,therefore, to increase the yield of the crop we need to identify
crop and weed by combining biomass sensing and digital image processing
technique. The biomass sensors like PH sensors and load cell sensors are used
to sense the biomass level in the farmland when the biomass level exceeds the
threshold value it automatically enables the camera. The camera captures the
image of  the particular farm land  where the higher biomass level indicated. The
captured image undergoes image processing technique. The intimation is done to farmers by sending
the weed present direction through SMS and  
posting the weed spotted image in the website.The proposed system detects weeds by color
using color detection algorithms. There are twenty kinds of weed that can be
detected using this system. The openCV  is
the interface used for color detection algorithm..There are twenty types of weeds that can be detected they
are Aeschynomene aspera L,
Aeschynomene indica L,
Amaranthus spinosus, Ludwigia adscendens, Ludwigia octovalvis, Acacia
farnesiana, Asclepias curassavica, Cassia alata, Cassia
obtusifolia, Cassia occidentalis, Cassia
occidentalis, Abutilon indicum (L.), Asteracantha longifolia, Biophytum sensitivum, Blumea oxyodonta DC, Cabomba aquatic, Cuscuta chinensis, Cuscuta reflexa, Flaveria trinervia. These above types of weeds are detected in digital
image processing part in the proposed system. This
proposed system aims to reduce the labor work spent on detecting the weed
manually, So the framers by using the system can spot the exact place where the
weed is present and also can know the weeds present direction through SMS.        ,                                                            
II.     
Methodology     The proposed system consist of two main
parts. first part is biomass sensing part where the proposed system uses
PHsensor and load cell sensor for detecting the biomass present in the farmland
. If the sensed value exceeds the thershold value then it enables the camera
for capturing the particular  part of
farm land. Second part is the digital image processing part in this part color
detection algorithms is used by interfacing openCV API, weed present in
captured image is detected using this algorithms.                Different stages in color detection algorithms are   discussed next.A.    RGB TO HSV ConverstionAfter the image is captured undergoes the RGB to HSV conversion where
the captured image is in the form of RGB should be converted to grey scale
image where HSV represents the hue , saturation and value of the particular
image.B.    Background SubractionBackground subtraction is a preprocessing step which is used for
extracting the particular object present in the image and subtracting the
background present. In the proposed system weeds picture will be extracted and
all other background image will be subtracted.C.    Morphological OperationA set of operations that process images based on
shapes. Morphological operations apply a structuring element to
an input image and generate an output image. The basic operation present are morphological errosion and
morphological dailationThe morphological erosion and dilation are used for Removing
noise, Isolation of individual elements and joining disparate elements in an
image, Finding of intensity bumps or holes in an image.                                
i.           
Morphological ErosionThe erosion
is a process applied to the image , it is similar to soil erosion it erodes
away the body of the foreground image. This method is used for removing small
white noise and detaching two connected objects.                               
ii.           
Morphological DilationIt is same as the opposite
effect of erosion. It will increase the size of the fore ground object. Erosion
when removing the white noise will also shrink the object, dilation is used to
dilate the object by increasing its size. The white noise won’t come back but
the object size will be increased. This process happens in dilation. D.    ContourThe contour is curve joining all continuous points which are having the
same intensity. The contour is used for shape analysis , object detection , In
our proposed system contour is used for setting the boundary and detecting the
object. E.    ImplementationThe Raspberry pi 3 model is used us the microcontroller. It is the
latest version of raspberry pi models. OpenCV is installed  and image processing codes are implemented
using python. If the values of PH sensor and piezoelectric sensor exceeds the
threshold values then the camera will be enabled. The camera captures the
particular farmland and image processing algorithm is implemented.  The weeds are spotted in the image and the
picture is posted in the website and intimation of weed detected direction is
made to farmers through SMS.  Microcontroller Raspberry piThe microcontroller used in the
proposed system is raspberry pi 3 version.  Result and
Discussion1.      
As it was mentioned
before , the image processing part is implemented using openCV .2.      
The goal is to find
the weeds present in rice field and the proposed system can detect twenty kinds
of weeds. Among the twenty kinds of weeds 18 kinds of weeds can be detected
using this proposed system, that is this yields 98% of classification accuracy.
There are 2 kinds of weed that have been wrongly classified isbecause of the
shades, the two kinds of weed doesn’t have any shades of yellow. Conclusion and discussion

Agriculture is one of the important
economical sector in tamilnadu. Now a day weed detection had become the most
interesting areas. This paper introduces the new weed detection system which is
capable of identifying twenty kinds of weed that occur in rice field. This
algorithm has a 98% classification accuracy is obtained.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

Post Author: admin

x

Hi!
I'm Erica!

Would you like to get a custom essay? How about receiving a customized one?

Check it out