Segmentation of Palm Oil Leaf Disease using

Zoning Feature Extraction

Ause Labellapansa, Ana Yulianti and Agus Yuliani

Department of Informatics,Universitas Islam Riau, Pekanbaru, Indonesia

Keywords:

Palm oil, Pests, Digital Image Processing, Zoning, Fitur Extraction.

Abstract:

Oil palm (Elaeis) is an industrial plant that produces large proﬁts, especially in Indonesia. One of the factors

that can affect the yield of this plant is destructive pests including Limacodidae and Psychidae. Delay in

dealing with pest problems can cause poor results. This study uses the help of digital image processing to

inentify two types of pests found on palm oil leaves of pests. Segmentation will be carried out to determine

the characteristics of Limacodidae and Psychidae pests. The image processing method used is the zoning

feature ecstasy. It is expected that knowing the types of pests suffered by oil palm trees can accelerate the

recovery of oil palm trees so as to produce good quality of fruit.

1 INTRODUCTION

Indonesia is the world’s largest Palm Oil producer.

It is spread out from Aceh region, the East Coast of

Sumatra, Java, Kalimantan and all the way to Su-

lawesi (Ermawati and Saptia, 2013). One of the fac-

tors that can affect the yield of palm oil is destructive

pests.

(Pribadi and Anggraeni, 2011) states that if plants

are in low humidity environment conditions, they will

be easily attacked by pests and diseases. This is sus-

pected due to saponin compounds found in plants

(which act as self-defense from insect attacks) will

decrease qualitatively and quantitatively so that the

plants will easily be harmed by pests.

Some of the destructive pests that attack the oil

palm plantations are Limacodidae and Psychidae.

The potential loss of yield caused by these two pests

can reach 35% (Wood et al., 1973). Limacodidae is a

palm-leaf-eating pest that often harms oil palm plan-

tations in North Sumatra.

The attack of the caterpillar pest which is a palm-

leaf-eating caterpillar has caused many problems.

This causes the loss of leaves of the plants which has

a direct impact on the decrease in production so this

indicates how serious the caterpillar attack is (Pahan,

2008).

To overcome this problem, computer system assis-

tance is needed by utilizing image processing knowl-

edge to identify these two types of pests. (Harahap

et al., 2018) identiﬁed oil palm leaf disease using the

Support Vector Machine method with an accuracy of

90% and (Aji et al., 2013) did the same thing using

artiﬁcial neural networks and produced an accuracy

value of 87.75%. Feature extraction for ﬁnding dis-

ease in leaves was carried out by (Arivazhagan et al.,

2013). The use of a deep convolution neural network

was carried out by (Sladojevic et al., 2016) to identify

13 leaf diseases with a precision level of 91% to 98%.

Detecting and classifying the plant leaf diseases based

by using GLCM and SVM on the Apple leaf has been

conducted by Sivakamasundari, G., & Seenivasagam

(Sivakamasundari and Seenivasagam, 2018) with ac-

curacy level about 92%. Our research is preliminary

research by identifying the image of palm oil leaves

and has not entered the classiﬁcation stage.

2 RESEARCH METHODOLOGY

The steps in this study are shown in Figure 1. The

image acquisition process is carried out by taking pic-

tures of leaves attacked by destructive pests. The im-

age will be processed from the original image to the

re-measurement stage by shrinking the pixel size to

600x250 pixels and followed by binery processing.

Zoning Feature Extraction will divide the leaf im-

age into several regions or zones of the same size, the

value of the features obtained from the method will

be used to determine the results of the image value of

palm oil leaves affected by Limacodidae and Psychi-

Labellapansa, A., Yulianti, A. and Yuliani, A.

Segmentation of Palm Oil Leaf Disease using Zoning Feature Extraction.

DOI: 10.5220/0009122100980101

In Proceedings of the Second International Conference on Science, Engineering and Technology (ICoSET 2019), pages 98-101

ISBN: 978-989-758-463-3

Figure 1: Research Scheme

dae pests. Zoning is one of the most popular methods

used for optical character document characterization

(Hegadi, 2012). The zoning calculation process is as

follows:

• Counting the number of black pixels per zone.

• Counting the zone that has the highest number of

pixels. Figure 2 and 3 is the number of Black Pix-

els in Each Zone affected by Limacodidae Image

and Psychidae

Figure 2: The Number of Each Zone for Limacodidae

Figure 3: The Number of Each Zone for Psychidae

• Calculating the feature value of each zone. The

feature values of each zone in the Limacodidae

affected are:

z1 =

= 0, 3 (1)

z2 =

= 0 (2)

z3 =

= 1 (3)

z4 =

= 0 (4)

While the feature values of each zone in the Psy-

chidae affected are:

z1 =

= 1 (5)

z2 =

= 0 (6)

z3 =

= 1 (7)

z4 =

= 0, 75 (8)

Figure 4: Image Causes by Limacodidae Pest

Figure 5: Image Causes by Psychidae Pest

Figure 6: Image from the Resizing process Caused by Li-

macodidae Pests

Figure 7: Image from the Resizing process Caused by Psy-

chidae Pests

Figure 4 is an image caused by the Limacodidae

Pest and ﬁgure 5 by the Psychidae Pest. The resizing

process is shown in Figures 6 and 7. The next stage,

the image of the leaf is changed to a grayscale image.

Then that grayscale image will undergo a process of

conversion to binaries using a threshold value. In this

technique, digital images will be classiﬁed into two

parts, namely objects and background.

Segmentation of Palm Oil Leaf Disease using Zoning Feature Extraction

The solution to the matrix of caterpillar impact

reference image with a threshold value is 122 for Li-

macodidae (Formula 9) and 88.5 for Psychidae (For-

mula 10)

f (x, y) =



255, i f f (x, y) ≥ 122

0, i f f (x, y) < 122

(9)

f (x, y) =



255, i f f (x, y) ≥ 88, 5

0, i f f (x, y) < 88, 5

(10)

Figure 8 is the image of the impact of Limacodi-

dae binary process and Figure 9 is the image of the

impact of Psychidae binary process.

Figure 8: Result of Binary Process by Limacodidae Pests

Figure 9: Result of Binary Process by Psychidae Pestss

The next step is to use the zone extraction feature

where the image of the leaf will be divided into sev-

eral regions or zones of the same size. The feature

values obtained from the method will be used to deter-

mine the results of the image values of palm oil leaves

affected by Limacodidae and Psychidae. Zoning is

one of the most popular methods used for document

optical characterization (Hegadi, 2012). The calcula-

tion process in the zoning method is as follows:

• Count the number of black pixels per zone.

• Calculates zones that have the highest number of

pixels.

• Calculates the feature value of each zone from the

feature value

3 RESULT AND DISCUSSION

Figure 10 is the result of zoning. Image 1 and 2 are

the images of Caterpillar Pests and image 3 and 4 are

Figure 10: Zoning Images

Figure 11: (Cont.) Zoning Images

images of Psychidae Pests. Each image will produce

4 regions.

Table 1 is the value data for Figure 10. The zona-

tion values of Figure 1 are 1, 0.40, 016, and 0. It can

be seen in Table 1 that there are signiﬁcant differences

in zones 2 and 3.

Table 1: This caption has one line so it is centered.

Images Value Class

Zone

Image 1 1 0,40 0,16 0

Lima

codidae

Image 2 0,92 0,96 0,28 1

Lima

codidae

Image 3 0,33 0,16 1 0

Psy

chidae

Image 4 1 0,17 0 0

Psy

chidae

ICoSET 2019 - The Second International Conference on Science, Engineering and Technology

100

4 CONCLUSION

In research that has been done by using zoning feature

extraction, values can be taken from each zone in the

image. The results of zoning can be developed into

the classiﬁcation stage using k-NN, SVM, or artiﬁcial

neural networks. The brightness, contrast, and back-

ground of the image greatly affect the results that will

be processed by the zoning feature extraction.

REFERENCES

Aji, A. F., Munajat, Q., Pratama, A. P., Kalamullah, H.,

Setiyawan, J., and Arymurthy, A. M. (2013). Detec-

tion of palm oil leaf disease with image processing and

neural network classiﬁcation on mobile device. Inter-

national Journal of Computer Theory and Engineer-

ing, 5(3):528.

Arivazhagan, S., Shebiah, R. N., Ananthi, S., and Varthini,

S. V. (2013). Detection of unhealthy region of plant

leaves and classiﬁcation of plant leaf diseases using

texture features. Agricultural Engineering Interna-

tional: CIGR Journal, 15(1):211–217.

Ermawati, T. and Saptia, Y. (2013). Kinerja ekspor minyak

kelapa sawit indonesia the export performance of in-

donesia’s palm oil. Buletin Ilmiah Litbang Perdagan-

gan, 7(2):129–148.

Harahap, L. A., Fajri, R. I., Syahputra, M. F., Rahmat,

R. F., and Nababan, E. B. (2018). Identiﬁkasi penyakit

daun tanaman kelapa sawit dengan teknologi image

processing menggunakan aplikasi support vector ma-

chine. In Talenta Conference Series: Agricultural and

Natural Resources (ANR), volume 1, pages 53–59.

Hegadi, R. S. (2012). Recognition of printed kannada nu-

merals based on zoning method. International Journal

of Computer Applications, 975:8878.

Pahan, I. (2008). Paduan Lengkap Kelapa Sawit. Niaga

Swadaya.

Pribadi, A. and Anggraeni, I. (2011). Pengaruh temper-

atur dan kelembaban terhadap tingkat kerusakan daun

jabon (anthocephalus cadamba) oleh arthrochista hila-

ralis. Jurnal Penelitian Hutan Tanaman, 8(1):1–7.

Sivakamasundari, G. and Seenivasagam, V. (2018). Classi-

ﬁcation of leaf diseases in apple using support vector

machine. International Journal of Advanced Research

in Computer Science, 9(1).

Sladojevic, S., Arsenovic, M., Anderla, A., Culibrk, D., and

Stefanovic, D. (2016). Deep neural networks based

recognition of plant diseases by leaf image classiﬁ-

cation. Computational intelligence and neuroscience,

2016.

Wood, B., Corley, R., Goh, K., et al. (1973). Studies on the

effect of pest damage on oil palm yield. Studies on the

effect of pest damage on oil palm yield., (58).

Segmentation of Palm Oil Leaf Disease using Zoning Feature Extraction

101