Integrated Intelligent Research (IIR) International Journal of Computing Algorithm

Volume: 05 Issue: 02 December 2016 Page No.84-87

ISSN: 2278-2397

Detection of Nutrient Deficiencies in Plant Leaves using

Image Processing
N. Minni1, N. Rehna2
1
Assistant Professor, Department of Computer Science, Avvaiyar Govt. College for women Karaikal, Puducherry, India
2
Department of Computer Science, SSS Shasun Jain College for women T.Nagar, Chennai, Tamil Nadu, India
Email: [email protected], [email protected]

Abstract—In this paper, we suggest a model for the automatic and cells development. It also increases protein production [2]
detection and classification of nutrient deficiencies in plant [9]. Potassium is also a necessary element for plants. It plays
leaves. In an agricultural country like India, farmers are facing the key role in biochemical and physiological functions. It
lot of problems in detecting the causes for the diseases in helps the cuticle to grow thicker. So it prevents from water
plants. Only when the causes are sorted out, the solution can loss and disease [1][9]. Sulfur is essential to maintain dark
be found to treat them. With naked-eye observation it is green color, kindle seed production, and stimulate root and
difficult to classify the deficiency present in leaves. So with the general plant growth and part of proteins, amino acids, and
help of image processing algorithms, we have proposed a vitamins important in respiration.[9] Calcium is a building
model to detect the type of deficiencies in the leaves. The color block of cell walls, and promotes root and leaf development
and texture features are used to recognize and classify the and activates enzymes involved in plant metabolism.[9]
deficiencies. The combinations of features prove to be very Magnesium is the building block of chlorophyll. It takes care
effective in deficiency detection. This paper presents an of the photosynthesis in plants.[9] So the deficient symptoms
effective method for detection of nutrient deficiencies in leaves for the macronutrients according to the literature are discussed
using color-texture analysis and k-means clustering. below.

Keywords - Macro nutrient, nutrient deficiency detection, B. Deficiency Symptoms

texture, clustering algorithm. 1) Nitrogen (N)
Nitrogen deficiency turns the green plants to become pale and
I. INTRODUCTION yellowish green. The stalks become weak. Symptoms of this
deficiency will occur on older and lower leaves. The leaves
Nutrients play a major role for the plants to grow healthy. The occurs as v-shaped in yellow color starting from the tip and
plants require all nutrients in the correct proportionate. These goes down the midrib [2]. It may have stunted growth and
nutrients help the plants to grow and reproduce in a healthy shorter internodes, small pale yellow leaves. Plant may be
way. If malnutrition exists in the plants , there will be light green and the older leaves are affected first.[3]
symptoms like unhealthy leaves. Malnutrition occurs if the
nutrients goes below or above the needed quantity. Plant 2) Phosphorus (P)
Nutrients are generally classified into macro and micro Phosphorus deficiency usually occurs on young plants. The
nutrients. Macro nutrients are those which are required in leaf tips become reddish purple and margins appear on older
relatively large amounts. Micro nutrients are needed in small leaves. Phosphorus deficient plants normally appear dark
amounts. The nutrients such as calcium, nitrogen, potassium, green.[2][5].They may have red or purplish color leaves,
magnesium, sulphur and phosphorus are the major nutrients. especially undersides and death of tissue or necrosis may
The elements such as boron, manganese, iron, copper, zinc, follow. [3]
chlorine are the micro nutrients. These micro nutrients are of
little importance.[1] 3) Potassium (K)
Potassium deficiency occurs on the lower leaves first. The
II. LITERATURE REVIEW symptoms are necrosis of leaf margins and they appear yellow.
[2] Yellow-streaked leaves, followed by browning and death of
Numerous techniques and methodologies are used in this area tips and margins. Older leaves are affected first[3]
is reviewed and briefly described here. Various papers had
been published for disease detection in plant leaves. But more 4) Sulfur (S)
research is yet to be done in the topic nutrient deficiency Sulphur deficiency symptoms occurs on small plants. The leafy
detection of leaves. Large number of researches is focused on parts of the tree appear yellow. The young leaves becoming
disease detection rather than deficiency of major nutrients [8]. yellow can be noticed with this deficiency. Generaly yellowing
In this paper only the macronutrients are analysed and their of the whole plant starts with the younger leaves and plants
deficiencies are identified. may be light green. Plants may be prevented from growing or
developing properly and exhibit delayed maturity. [3]
A. Need for the nutrients
One of the component of nitrogen is the chlorophyll. So 5) Calcium (Ca)
nitrogen provides green colour to the leaves. It helps to Calcium deficiency occurs in plants like corn. The tips of the
improve the leaf crops quality and also to bear healthy fruits. If leaves get stuck to the next lower leaf and appears one after
this nutrient is deficient , it affects the overall growth of the another like a ladder.[2] The older, larger leaves just above the
crop. The leaves colour turns from light green to yellow[2][9]. bottommost ones will show the first symptoms. Yellow/brown
Phosphorus another important nutrient plays the role in root
84
Integrated Intelligent Research (IIR) International Journal of Computing Algorithm
Volume: 05 Issue: 02 December 2016 Page No.84-87
ISSN: 2278-2397
spots occur, which are surrounded often by a sharp brown zero[1]. The green pixels are masked so that they need not be
outlined edge. taken for further analysis.

6) Magnesium (Mg) RGB IMAGE ACQUISITION

Magnesium deficiency appears in the lower leaves. These
leaves are yellowish and white near the interveins. [2] There
will be yellow marking spots between veins of older leaves
while veins remain green. Yellow areas may turn brown and
then die. Leaves look burnt from the tips. Yellowing also may RGB TO HSI TRANSFORMATION
occur on older leaves. [3] Instead of mixing all different
techniques at once, color histograms have been used to retrieve
the major color contents of the image. Texture segmentation is
used to identify the desired markings or spots based on the IMAGE SEGMENTATION
texture configurations of an image. Texture analysis is
analyzing the image by their texture content. The image
regions are characterized and analyzed by the texture features
such as their roughness, silkyness etc.[7]
FEATURE EXTRACTION
The texture analysis uses the algorithm named co-occurrence
matrix method [7] and K-means Clustering method. The
algorithm K-means clustering attempt to classify pixels based
on a set of features into K number of classes. The sum of
squares of distances between the objects has been minimized CREATE TESTING DATABASE
and the corresponding cluster or class centroid for doing
classification. However, in this work K-means clustering has
been used and the leaf image is separated into four clusters in
which one or more clusters hold the disease. [7]
SEPARATE NORMAL FROM
III. PROPOSED METHODOLOGY DEFICIENT LEAVES

In our proposed approach, to detect the nutrient deficiencies in

leaves the properties such as colouring and texture of leaves
must be an analysed. So the properties for nutrient deficiencies
are stored in a database called training database. The IDENTIFY THE DEFICIENCIES
properties of leaves which have to undergo the deficiency FOR THE DEFICIENT LEAVES
detection process are stored in a database called testing
database. A methodology called Nutrient Deficiency Detection
using image processing have been proposed. The framework
of the proposed method is shown in Fig1. Fig. 1. Nutrient Deficiency detection Methodology
A. RGB image acquisition
2) Extract useful segments: The infected portion of the
Initially the images of various types of leaves are captured
using a digital camera. leaves are then segmented into patches of equal size. The size
of the patch is chosen such that the significant information is
B. RGB to HSI transformation not lost. All segments do not have useful information. So the
The RGB images of the leaves are represented in Hue Intensity patches having more than 70 percent of significant information
Saturation (HIS)color model. The HIS color space are considered for further analysis. For segmenting the
representation is widely accepted as it is related to human essential parts of segments such as shape of leaf tips, interveins
perception[13]. The Hue component refers to the dominant , region between interveins, spots etc statistical region merging
color of the image. Saturation gives the corresponding purity. algorithm is used. [6]
It is the amount of white light added to the hue value. Intensity
means the amplitude of the light present [12]. So in this paper D. Feature Extraction
the color space is transformed from RGB to HSI The main features extracted for consideration are: Texture,
representation. After the transformation, the H component is Color
taken into account for further analysis. 1) Texture: The texture of leaves are extracted using
statistical region merging . They may include features such as
C. Image segmentation death of tips and margin, curling, reddish purple tip etc. These
This step is divided into two parts features are stored in the testing database.

1) Masking green pixels: The pixels which are mostly 2) Color separation: The HSI (Hue, Saturation and
green are identified first. Then a threshold value is calculated Intensity) color model is used for representing the colors of the
such that if the green part of the pixel intensity is less than images. The color histogram represents the distribution of
the computed threshold value, those pixels are to be masked. color in images. Here we compute the intensity of the colors
The red, green and blue components of these pixels are made present in leaves and represent them in a histogram[10][12].
85
Integrated Intelligent Research (IIR) International Journal of Computing Algorithm
Volume: 05 Issue: 02 December 2016 Page No.84-87
ISSN: 2278-2397
Also, the intensity of each color is stored in testing database in
percentage. Table II. Testing Database - Samples Taken

E. Creating Testing database

The features of the leaves extracted in the previous phase are
stored in the testing database. For the proposed method the
features such as types of leaves and color of the leaves in
appropriate regions and the difference in their texture are
identified for each nutrient from the literature studied. They are
recorded in the testing database for comparison with the
experimental leaves.

F. Separate Normal and Deficient leaves.

The K means clustering algorithm[11] is used for clustering.
Those leaves with maximum pixels masked are considered to
be normal. The others which have values in the histogram are
classified as deficient. Here clusters will be classified in two
classes. Normal, Deficient

G. Identify the deficiencies for the deficient leaves

Only the deficient clusters from the previous step are taken as
input for this step. Here K-NN algorithm is applied to identify
the deficiencies[11]. The features of leaves in the testing
database are compared against the training database. If
maximum features match with the features of a particular
deficiency, then they are affected by that deficiency. Using
this technique, the clusters are formed for the different
deficiencies. Hence, finally the leaves are classified and fall
into either one of four groups (ie) potassium, calcium, nitrogen,
magnesium.
IV. RESULTS AND DISCUSSION

The training database consisting of the nutrients and their

deficiency symptoms is created as in table 1. [1][2][3][4][9]
The samples taken for deficiency detection of nutrients is
given in Table II. The sample leaves are converted into their Table III. Testing database – Extraction of colors using hue,
hue, saturation, intensity images. The colors present are saturation, intensity images
extracted from these images and recorded in the testing
database. Table III shows the hue, saturation and intensity
components of the samples. The Hue component is taken for
further analysis.

Table 1: Training database

Deficiency Symptoms
Nutrient Type of
Color Region Texture
leaves
Yellow / Death of
Calcium New Spots
brown leaf tips
Nitrogen Old Pale yellow Whole
Curling
Edge/
Pottasium Old Brown of leaf
segments
tips
Leaf tips
Between
Magnesium Old Yellow look
leaf veins
burnt
Sulphur New Yellow Whole
Leaf tips
Reddish Leaf tips &
Phosporus Old look
purple margins
burnt

The Hue components computes the colors present in the leaves

in percentage. Using this value and comparing with training
database, the appropriate nutrient which is deficient in the TABLE IV. Deficient nutrients for the samples taken
leaves is found. The following Table IV lists the samples and
their deficient nutrients found using the proposed approach. Samples English Name Nutrient Deficient
Sample 1 Beans Calcium

86
Integrated Intelligent Research (IIR) International Journal of Computing Algorithm
Volume: 05 Issue: 02 December 2016 Page No.84-87
ISSN: 2278-2397
Sample 2 Custard Apple Magnesium
Sample 3 Hibiscus Phosphorus
Sample 4 Ixora Nitrogen
Sample 5 Jasmine Healthy leaf
Sample 6 Hibiscus Sulphur
Sample 7 Rose Potassium
Sample 8 Sapodilla Healthy leaf

V. CONCLUSION

Nutrient Deficiency detection is a system which identifies the

unhealthy leaves and the nutrients which are deficient in them.
This makes use of the image processing technique. Colors and
texture of leaves are analysed for detection. The proposed
system consists of preprocessing, feature extraction,
segmentation , training and classification and finally
identification. This paper proposes a valuable approach which
supports the accurate detection of nutrients which are deficient
in the plant leaves. This work can further be extended
focussing on texture analysis. More efficient algorithms
including more features of the texture can be developed.

References
[1] Guide to Symptoms of Plant Nutrient
Deficiencies ,Shanyn Hosier,Instructional Specialist Urban
Horticulture Lucy Bradley, Agent Urban Horticulture
[2] Nutrient deficiencies row crops ISU.pdf, IOWA state
university, extension and outreach
[3] Nutrient Deficiency Symptoms Barbara J. Bromley,
Mercer County Horticulturist 10
[4] John Sawyer, Department of Agronomy. “Nutrient
Deficiencies and Application Injuries in Field Crops” in
IOWA State university (University Extension) IPM 42
Revised July 2004.
[5] Disease and Nutrient Deficiency Detection in Cotton Plant
Ashish Miyatra, Sheetal Solanki ijedr, Volume 2, Issue 2,
2014
[6] Statistical Region Merging Richard Nock and Frank
Nielsen, IEEE transactions on pattern analysis and
machine intelligence, vol. 26, no. 11, November 2004
[7] Remote Area Plant Disease Detection Using Image
ProcessingSabah Bashir, Navdeep Sharma IOSR Journal
of Electronics and Communication Engineering
(IOSRJECE) Sep-Oct2012
[8] A Survey on Disease and Nutrient Deficiency Detection in
Cotton PlantAshish Miyatra , Dulari Bosamiya, Nilesh
Kamariya IJRITCC | November 2013
[9] Nutrient deficiencies in export tree and food crops:
literature review and field observations Article · January
2000 Alfred E. Hartemink and R. Michael Bourke
[10] A Review of Image Processing for Pomegranate Disease
Detection Manisha A. Bhange, Prof. H.A. Hingoliwala,
International Journal of Computer Science and
Information Technologies, 2015
[11] A Review paper on : Agricultural Plant Leaf Disease
Detection Using Image Processing Prof. Atul
Shire,Prof.Umesh Jawarkar, Mr. Manoj Manmode ,
International Journal of Innovative Science, Engineering
& Technology jan 2015
[12] Automatic detection and classification of plant disease
through image processing, Mr. Pramod S. landge, Sushil
A. Patil, Dhanashree S. Khot,Omkar D. Otari, Utkarsha G.
Malavkar IJARCSSE 2013
[13] Gonzalez, R., R.E. Woods. 2008 Digital Image processing.
Third edition , Pearson Education, Prentice-Hall, Inc.

87