2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

CIELuv Color Space for Identification and

Segmentation of Disease Affected Plant Leaves
Using Fuzzy based Approach
Ganesan P1* G.Sajiv2, Megalan Leo.L3
1 2, 3
Dept. of ECE, Department of Electronics and Department of Electronics and
Vidya Jyothi Institute of Telecommunication Engineering, Telecommunication Engineering,
Technology, Sathyabama University, Chennai, Sathyabama University, Chennai,
Hyderabad, India. India. India.
[email protected], [email protected], [email protected]

Abstract— Economy of developing countries identification of plant disease is advantageous as it

mainly depends on the agriculture productivity. The detects very early stage of symptoms of plant
identification and control of plant disease plays a vital diseases and also it lessens the burden of workers in
role in improving the quality and quantity of monitoring the large fields. Image processing
agriculture productivity. It is very difficult to monitor
techniques can be applied to solve the problem of
or detect the crop disease using the traditional manual
methods. Computer based automated techniques for the identification and extraction of the affected parts of
identification of plant disease is advantageous as it plants in efficient manner. In this work, fuzzy based
detects very early stage of symptoms of plant diseases segmentation method for the early identification and
and also it lessens the burden of workers in monitoring segmentation of disease affected plant leaves is
the large fields. Image processing techniques can be proposed.
applied to solve the problem of identification and Computer vision based segmentation
extraction of the affected parts of plants in efficient method presents the accurate and rapid solutions of
manner. In this work, fuzzy based method for the early detection and control of plant leaf diseases as well as
identification and segmentation of disease affected plant
improve the crop productivity. Image segmentation
leaves is proposed.
can be applied to extract the diseased part of the plant
Keywords— segmentation; MFCM; color space; leaf from the input image.
image enhancement; plant leaves; Image processing plays a vital role for
detecting and analyzing and plant leaf diseases. The
image processing can be applied not only for the
I. INTRODUCTION detection of plant leaf disease but also used for many
In the field of agriculture, an automated, agricultural purposes such as
computerized technique for the detection of disease 1. To identify diseased leaf, fruit, or stem.
affected plant leaf is a hot topic because it can 2. To measure the disease level.
automatically monitor, detect and control plant 3. To compute the size and shape of fruits disease
diseases almost immediately as become visible on affected area.
plants. Disease on plant leaves cause major 4. To find out color of fruit or disease affected
production and economic losses in agricultural area
industry. Identification and classification of disease The skeleton of paper is framed as follows.
harshness is vital factor to reduce the economic loss An overview of literature survey is presented in
and for the excellent use of pesticide to control the section 2. The section 3 discusses the robotics for
disease. The economy of developing countries mainly warehouse management system. The proposed
depends on the agriculture productivity. The approach for warehouse management system using
identification and control of plant disease plays a microprocessor based mobile robotic system is
vital role in improving the quality and quantity of explained in section 4. Finally the section 5
agriculture productivity. concludes the paper.
It is very difficult to monitor or detect the
crop disease using the traditional manual methods.
Computer based automated techniques for the
889
978-1-5090-4855-7/17/$31.00 ©2017 IEEE
 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

II. COLOR SPACE FOR SEGMENTATION AND  ൌ ͲǤͲͳͻ͵͵Ͷ ‫  כ‬൅ ͲǤͳͻͳͻ͵ ‫

כ‬൅ ͲǤͻͷͲʹʹ͹ ‫ כ‬ሺ͵ሻ
IDENTIFICATION
In color representation system, color model The conversion of CIEXYZ to CIELUV is given as
can be explained as a geometric demonstration of
color information based on dissimilar color ଵൗ
ଢ଼ ଷ ଢ଼
components. RGB color system space is a  ൌ ͳͳ͸ ቀ ቁ െ ͳ͸‹ˆ ൐ ͲǤͲͲͺͺͷ͸ (4)
ଢ଼బ ଢ଼బ
fundamental color space wherein red, green and blue ଢ଼
 ൌ ͻͲ͵Ǥ͵ ቀ ቁ ‹ˆ
ଢ଼
൑ ͲǤͲͲͺͺͷ͸(5)
colors are depicted as shown in figure 1. ଢ଼బ ଢ଼బ
— ൌ ͳ͵ሾሺǡ ǡ ሻ െ ሺ଴ ǡ ଴ ǡ ଴ ሻ] (6)
˜ ൌ ͳ͵ሾሺǡ ǡ ሻ െ ሺ଴ ǡ ଴ ǡ ଴ ሻ] (7)

III. MODIFIED FUZZY BASED SEGMENTATION

The modified FCM (MFCM) is

elucidated as how it minimizing its objective function
as presented in (8)

୫ ሺǡ ሻ ൌ  σୡ୧ୀଵ σ୬୩ୀଵ Ɋ୫

୧୩ ԡš ୩ െ ˜୧ ԡ
ଶ
(8)
Fig. 1. RGB Color Space The cluster centers (9) and membership (10)
computed as
The salient features of CIELuv color space is as
follows σ౤ ౣ
ౡసభ ஜ౟ౡ ଡ଼ౡ
୧ ൌ σ౤ ౣ (9)
• Perceptually uniform ౡసభ ஜ౟ౡ
ଶൗ ିଵ
• Device independent ୡ ԡ୶ౡ ି୴౟ ԡ ሺ୫ିଵሻ
Ɋ୧୩ ൌ ቊσ୨ୀଵ ൜ฮ୶ ି୴ ฮൠ ቋ  (10)
• Linear ౡ ౠ

• Intuitive The spatial information is calculated as

• Color information (red/blue and
σ ୙౟ౡ ஑ౡమ
yellow/blue) are stored in u and v ౡ‫א‬ቀ౔ౠ ቁ
୧୨ ൌ σ୩‫א‬୛൫ଡ଼ౠ ൯ ୧୩ Ƚ୩ଵ ൅ σౙ (11)
components. ౪సభ σౡ‫౓א‬ቀ౔ ቁ ୙౪ౡ
ౠ

The weightሺ୨୧ ሻ of pixels in each cluster is

computed as
ଵ
୨୧ ൌ మ (12)
ቛ౔ౠ ష౒౟ ቛ
ష൞ మ ౙ ൢ
σ౤ ቛ౔ౠ ష౒౟ ቛ ቀ ቁ
ଵାୣ ౠసభ ౤
The modified objective function is represented
as given in (13)
୫ ୫
 ୑୭ୢୀ σ୬୩ୀଵ σୡ୧ୀଵ൫୧୩ ୨୧ ൯ԡ୩ െ ୧ ԡଶ (13)

IV. PROPOSED APPROACH FOR SEGMENTATION

AND IDENTIFICATION OF DISEASE
AFFECTED PLANT LEAVES
Fig. 2.CIELuv RGB Color Space Proposed approach for the segmentation and
identification of disease affected plant leaves is
The transformation equations for CIEXYZ is
illustrated in fig 3. The algorithm for the
represented as
segmentation of segmentation and identification of
 ൌ ͲǤͶͳʹͶͷ͵ ‫  כ‬൅ ͲǤ͵ͷ͹ͷͺ ‫
כ‬൅ ͲǤͳͺͲͶʹ͵ ‫ כ‬ሺͳሻ disease affected plant leaves is explained as follows.
 ൌ ͲǤʹͳʹ͸͹ͳ ‫  כ‬൅ ͲǤ͹ͳͷͳ͸ ‫
כ‬൅ ͲǤͲ͹ʹͳ͸ͻ ‫ כ‬ሺʹሻ

890
978-1-5090-4855-7/17/$31.00 ©2017 IEEE
 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

1. Collect the input image from any image

acquisition device such as digital camera. 6. Based on the segmentation result, the
disease affected portion is extracted.
2. Unnecessary information such as noise
should be removed from image using 7. The disease affected portion of the leaf
median filter of kernel size 3. image is analyzed using image quality
parameters.
3. Most of the image acquisition devices
present the input image which is usually in
RGB color space. The device dependent V. RESULT AND DISCUSSION
RGB color space is not perceptually
uniform. SO it is necessary to transform the Upshot of the proposed approach for the
input image from RGB to CIELuv color segmentation of disease affected portion of the leaf
space. image is depicted in figure 4. Here original input
image is divided into number of clusters as
4. Transformed CIELuv image should be unaffected or affected portion of the plant leaf.
enhanced by sharpening the fine details of Analysis of disease affected plant leaves using
the image. various image quality parameters are presented in
table 1. Figure 5 illustrates the segmentation of
5. Modified FCM is applied on the image to disease affected plant leaves using other methods.
segment it into number of clusters.

Image Image Pre Color Image

Acquisition Processing Transformation Enhancement

Image Feature Fuzzy

Analysis Extraction Segmentation

Fig. 3. Proposed Approach

TABLE I. ANALYSIS OF DISEASE AFFECTED PLANT LEAVES

Root Mean Square Error 37.780 37.742

Variance 1276.1 1.3282
Mean 23.595 10.190
Standard Deviation 37.780 37.742
Entropy 3.4399 1.1579
Correlation 0.8620 0.8566
Skewness 1.7292 4.1442
Smoothness 0.9999 1.0000
Contrast 0.3063 0.3605
Root Mean Square Error 37.780 37.742
Variance 1276.1 1.3282
Mean 23.595 10.190
Standard Deviation 37.780 37.742
Entropy 3.4399 1.1579
Correlation 0.8620 0.8566
Skewness 1.7292 4.1442
Smoothness 0.9999 1.0000
Contrast 0.3063 0.3605
891
978-1-5090-4855-7/17/$31.00 ©2017 IEEE
 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

Fig. 4. Outcome of the proposed approach

proposed method is analyzed using image quality

VI. CONCLUSION
parameters to show the efficiency for an early
In this paper, a new method for the extraction detection and segmentation of disease affected plant
of the disease affected region of plant leaf images is leaves. Though the experimental results proved the
proposed. The proposed color based method is robustness and efficiency of the proposed method,
realized by means of CIELuv color space and the CIELuv color space can be replaced by the device
Modified fuzzy C-means clustering algorithm. independent, perceptually uniform and linear color
Though many images had experimented using the space like CIELAB, CIELch or HSI for further
proposed method, the outcome of only one image is improvement.
demonstrated and explained in this manuscript. The

892
978-1-5090-4855-7/17/$31.00 ©2017 IEEE
 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

Fig. 5. Segmentation of disease affected plant leaves using other methods

893
978-1-5090-4855-7/17/$31.00 ©2017 IEEE
 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM)

[11] Y. Hao, L. Changshun and P. Lei, “An improved

REFERENCES method of image edge detection based on wavelet
transform,” Computer Science and Automation
[1] Gupta, M.; Tazi, S.N.; Jain, A.; Deepika, "Edge Enginnering (CSAE), pp. 678-681, June 2011.
Detection Using Modified Firefly Algorithm," [12] Gautam, D.; Ahmed, M., "Efficient fuzzy edge
in Computational Intelligence and Communication detection using successive Otsu's method,"
Networks (CICN), 2014 International onference on , in Convergence of Technology (I2CT), 2014
vol., no., pp.167-173, 14-16 Nov. 2014 International Conference for , vol., no., pp.1-5, 6-8
[2] P. Ganesan and V. Rajini, "A method to segment color April 2014
images based on modified Fuzzy-Possibilistic-C- [13] P. Ganesan, V. Rajini, B. S. Sathish and K. B. Shaik,
Means clustering algorithm," Recent Advances in "HSV color space based segmentation of region of
Space Technology Services and Climate Change 2010 interest in satellite images," 2014 International
(RSTS & CC-2010), Chennai, 2010, pp. 157-163. Conference on Control, Instrumentation,
doi: 10.1109/RSTSCC.2010.5712837 Communication and Computational Technologies
[3] Wenlong Fu; Johnston, M.; Mengjie Zhang, "Low- (ICCICCT), Kanyakumari, 2014, pp. 101-105.
Level Feature Extraction for Edge Detection Using doi: 10.1109/ICCICCT.2014.6992938
Genetic Programming," in Cybernetics, IEEE [14] X. Yang, D Qi and X. Li, “Multi-scale edge detection
Transactions on , vol.44, no.8, pp.1459- of wood defect images based on the dyadic wavelet
1472,Aug.2014 transform,” Machine Vision and Human-Machine
Interface(MVHI), Kaifeng, China, pp. 120-123, April
[4] Wenshuo Gao; Xiaoguang Zhang; Lei Yang; 2010
Huizhong Liu, "An improved Sobel edge detection," [15] Di Jia; Cheng-long Xiao; Jin-guang Sun, "Edge
Computer Science and Information Technology detection method of Gaussian block distance,"
(ICCSIT), 2010 3rd IEEE International Conference on in Image Processing (ICIP), 2015 IEEE International
, vol.5, no 1 Conference on , vol., no., pp.3049-3053, 27-30 Sept.
2015
[5] Kiani, A.; Sahebi, M.R., "Edge detection based on the
Shannon Entropy by piecewise thresholding on remote [16] W. Liang, C. Da and L. Dun, “A single threshold edge
detection algorithm based on multi scale fusion,”
sensing images," in Computer Vision, IET , vol.9,
Computational Intelligence and Security, Guangzhou,
no.5,pp.758-768,2015. vol. 2, pp. 1710-1715, Nov. 2006
[17] Anand, A.; Tripathy, S.S.; Kumar, R.S., "An improved
[6] Ganesan, P., and V. Rajini. " Segmentation and
denoising of noisy satellite images based on modified edge detection using morphological Laplacian of
fuzzy c means clustering and discrete wavelet Gaussian operator," in Signal Processing and
Integrated Networks (SPIN), 2015 2nd International
transform for information retrieval”, International
Conference on , vol., no., pp.532-536, 19-20 Feb.
journal of engineering and technology, vol.5, no.5,
2015
2013, pp.3856-3869.
[18] Kumar, K.; Mustafa, N.; Jian-Ping Li; Shaikh, R.A.;
[7] Melin, P.; Gonzalez, C.I.; Castro, J.R.; Mendoza, O.;
Khan, S.A.; Khan, A., "Image edge detection scheme
Castillo, O., "Edge-Detection Method for Image
Processing Based on Generalized Type-2 Fuzzy using wavelet transform," in Wavelet Active Media
Logic," in Fuzzy Systems, IEEE Transactions on , Technology and Information Processing
(ICCWAMTIP), 2014 11th International Computer
vol.22, no.6, pp.1515-1525, Dec.2014.
Conference on , vol., no., pp.261-265, 19-21 Dec.
[8] Caixia Deng; Weifeng Ma; Yin Yin, "An edge 2014
detection approach of image fusion based on improved [19] Sajiv, G.,Ganesan, P,” Comparative study of
Sobel operator," Image and Signal Processing (CISP), possiblistic fuzzy c-means clustering based image
2011 4th International Congress on , vol.3, no., segmentation in RGB and CIELuv color space”,
pp.1189,1193, 15-17 Oct. 2011. International Journal of Pharmacy and
[9] Shaik, K.B.,Ganesan, P.,Kalist, V.,Sathish, Technology,March-2016,Vol. 8, Issue No.1,pp.
B.S.,Jenitha, J.M.M,” Comparative Study of Skin
Color Detection and Segmentation in HSV and YCbCr
Color Space “, Procedia Computer Science, 57 ( 2015
), pp. 41 – 48
[10] Shiling Sun; Shuxun Yan; Ying Wang; Yun Li,
"Medical CT Edge Detection Algorithm Based on
Improved Fuzzy Clustering Analysis," in Intelligent
Systems Design and Engineering Applications
(ISDEA), 2014 Fifth International Conference on ,
vol., no., pp.607-610, 15-16 June 2014

894
978-1-5090-4855-7/17/$31.00 ©2017 IEEE