PENGARUH KOMBINASI HISTOGRAM OF ORIENTED GRADIENTS DAN FITUR WARNA LAB TERHADAP SUPPORT VECTOR MACHINE DALAM MENDETEKSI PARASIT MALARIA PADA CITRA APUSAN DARAH

Mardhiyah, Siti (2024) PENGARUH KOMBINASI HISTOGRAM OF ORIENTED GRADIENTS DAN FITUR WARNA LAB TERHADAP SUPPORT VECTOR MACHINE DALAM MENDETEKSI PARASIT MALARIA PADA CITRA APUSAN DARAH. Other thesis, UPN Veteran Yogyakarta.

	Text ABSTRAK.pdf Download (49kB)
	Text COVER.pdf Download (167kB)
	Text DAFTAR ISI.pdf Download (173kB)
	Text DAFTAR PUSTAKA.pdf Download (183kB)
	Text PENGESAHAN PEMBIMBING.pdf Download (253kB)
	Text SKRIPSI FULL_SITI MARDHIYAH.pdf Restricted to Repository staff only Download (3MB)

Abstract

vi
ABSTRAK
Terdapat berbagai macam algoritma untuk melakukan klasifikasi citra, salah satunya
adalah Support Vector Machine (SVM). Support Vector Machine efektif dalam mengolah
data berukuran kecil dan mampu menghasilkan akurasi tinggi, selain itu juga memiliki waktu
komputasi yang efisien. Namun Support Vector Machine membutuhkan metode ekstraksi
fitur untuk mendapatkan informasi spesifik dari citra yang digunakan dalam proses
klasifikasi. Penggunaan fitur tunggal masih belum mampu memenuhi kebutuhan informasi
tersebut, sehingga performa Support Vector Machine kurang maksimal jika dibandingkan
dengan metode klasifikasi lain. Disisi lain, dengan menggunakan beberapa fitur, akurasi
yang diperoleh menjadi lebih baik dibandingkan penggunaan fitur tunggal. Penelitian ini
melakukan analisis pengaruh penggunaan gabungan Histogram of Oriented Gradients dan
fitur warna Lab terhadap performa dan waktu proses Support Vector Machine dalam
mendeteksi parasit malaria pada citra apusan darah dengan membandingkannya terhadap
Support Vector Machine yang hanya menggunakan Histogram of Oriented Gradients.
Hasil penelitian dalam mendeteksi parasit malaria pada 2000 citra, yang terdiri dari
1000 citra yang terinfeksi malaria dan 1000 citra apusan darah yang tidak terinfeksi malaria,
dengan rasio pembagian data training dan testing 80:20, serta melalui preprocessing,
ekstraksi fitur, dan normalisasi fitur, menunjukkan bahwa Support Vector Machine dengan
gabungan Histogram of Oriented Gradients dan fitur warna Lab memiliki performa yang
lebih baik dengan perbedaan waktu proses yang tidak signifikan dibandingkan dengan
Support Vector Machine yang hanya menggunakan Histogram of Oriented Gradients. Dari
hasil cross-validation dan pemilihan parameter terbaik, didapatkan bahwa kombinasi
parameter dengan k=6, C=3, dan gamma=0,002 memiliki performa tertinggi. Pada parameter
tersebut, Support Vector Machine dengan gabungan Histogram of Oriented Gradients dan
fitur warna Lab memperoleh akurasi, precision, recall, dan f1-score sebesar 93,25%.
Sebaliknya, Support Vector Machine yang hanya menggunakan ekstraksi Histogram of
Oriented Gradients mendapatkan akurasi sebesar 89%, precision sebesar 89,10%, serta
recall dan f1-score sebesar 89,99%.
Kata kunci: deteksi malaria, citra apusan darah, Histogram of Oriented Gradients, fitur
warna Lab, Support Vector Machine
vii
ABSTRACT
There are various algorithms for image classification, one of which is Support Vector
Machine (SVM). Support Vector Machine is effective in handling small-sized data and can
produce high accuracy, as well as having efficient computation time. However, Support
Vector Machine requires feature extraction methods to obtain specific information from the
images used in the classification process. The use of a single feature is still insufficient to
meet these information needs, resulting in suboptimal performance of Support Vector
Machine compared to other classification methods. On the other hand, using multiple
features can achieve better accuracy than using a single feature. This research analyzes the
impact of using a combination of Histogram of Oriented Gradients and Lab color features
on the performance and processing time of Support Vector Machine in detecting malaria
parasites in blood smear images by comparing it to Support Vector Machine that only uses
Histogram of Oriented Gradients.
The research results in detecting malaria parasites on 2000 images, consisting of
1000 images infected with malaria and 1000 blood smear images not infected with malaria,
with a data training and testing split ratio of 80:20, as well as through preprocessing, feature
extraction, and feature normalization, show that Support Vector Machine with a
combination of Histogram of Oriented Gradients and Lab color features has better
performance with no significant difference in processing time compared to Support Vector
Machine that only uses Histogram of Oriented Gradients. From the results of cross-
validation and the selection of the best parameters, it was found that the parameter
combination with k=6, C=3, and gamma=0.002 had the highest performance. With these
parameters, Support Vector Machine with a combination of Histogram of Oriented
Gradients and Lab color features achieved an accuracy, precision, recall, and f1-score of
93.25%. In contrast, Support Vector Machine that only uses Histogram of Oriented
Gradients achieved an accuracy of 89%, a precision of 89.10%, and both recall and f1-score
of 89.99%.
Keywords: malaria detection, blood smear images, Histogram of Oriented Gradients, Lab
color features, Support Vector Machine

Item Type:	Thesis (Other)
Uncontrolled Keywords:	malaria detection, blood smear images, Histogram of Oriented Gradients, Lab color features, Support Vector Machine
Subjects:	Z Bibliography. Library Science. Information Resources > ZA Information resources
Divisions:	Faculty of Engineering, Science and Mathematics > School of Engineering Sciences
Depositing User:	Eko Yuli
Date Deposited:	24 Jul 2024 06:57
Last Modified:	24 Jul 2024 06:57
URI:	http://eprints.upnyk.ac.id/id/eprint/40395

Actions (login required)

View Item