Pengaruh Peningkatan Resolusi Citra pada Akurasi Klasifikasi Kualitas Biji Jagung Menggunakan Ekstraksi Fitur Histogram dan Metode K-Nearest Neighbor

Kumalahadi, Putri Ayu (2025) Pengaruh Peningkatan Resolusi Citra pada Akurasi Klasifikasi Kualitas Biji Jagung Menggunakan Ekstraksi Fitur Histogram dan Metode K-Nearest Neighbor. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5001211071-Undergraduate_Theses.pdf Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Jagung (Zea mays L.) merupakan komoditas pertanian penting di Indonesia, di mana kualitas bijinya sangat menentukan nilai jual. Proses klasifikasi kualitas biji jagung secara manual masih umum dilakukan dan cenderung memakan waktu, subjektif, serta kurang konsisten. Penelitian ini bertujuan untuk mengembangkan sistem klasifikasi kualitas biji jagung berbasis citra digital menggunakan metode K-Nearest Neighbor (KNN) dengan ekstraksi fitur histogram. Penelitian ini juga mengevaluasi pengaruh peningkatan resolusi citra sebesar dua kali dan tiga kali lipat terhadap akurasi klasifikasi. Dalam penelitian ini, sebanyak 200 citra biji jagung digunakan, terdiri dari 120 citra latih dan 80 citra uji yang terbagi merata ke dalam empat kategori, yaitu biji busuk, berjamur, normal, dan rusak. Tahapan pemrosesan citra mencakup cropping ke ukuran 200200 piksel, peningkatan resolusi citra, konversi ke grayscale, filtering dengan median filter 99, dan Adaptive Histogram Equalization. Fitur yang diekstraksi meliputi lima parameter histogram, yaitu mean, standar deviasi, skewness, kurtosis, dan entropy. Hasil klasifikasi menunjukkan bahwa peningkatan resolusi citra berpengaruh positif terhadap akurasi. Akurasi tertinggi diperoleh pada resolusi dua kali, yaitu 90.83% pada citra latih dan 70% pada citra uji, dibandingkan dengan resolusi tiga kali (90% latih, 66.25% uji) dan resolusi asli (85.83% latih, 67.5% uji). Metode KNN juga menunjukkan performa yang lebih baik dibandingkan metode Naïve Bayes Classifier dari penelitian sebelumnya. Dengan demikian, peningkatan resolusi citra dan pemanfaatan metode KNN terbukti efektif dalam meningkatkan akurasi klasifikasi kualitas biji jagung.
=================================================================================================================================
Corn (Zea mays L.) is an important agricultural commodity in Indonesia, where the quality of its kernels significantly affects its market value. The manual classification of corn kernel quality is still commonly practiced, but it is often time-consuming, subjective, and inconsistent. This study aims to develop a corn kernel quality classification system based on digital image processing using the K-Nearest Neighbor (KNN) method with histogram feature extraction. It also evaluates the effect of image resolution enhancement, by two and three times, on classification accuracy. In this research, 200 corn kernel images were used, consisting of 120 training images and 80 testing images, evenly distributed into four categories, rotten, moldy, normal, and damaged kernels. Image preprocessing steps included cropping to a uniform size of 200200 pixels, resolution enhancement, grayscale conversion, 99 median filtering, and Adaptive Histogram Equalization. Five histogram-based features were extracted, mean, standard deviation, skewness, kurtosis, and entropy. The classification results show that resolution enhancement positively impacts accuracy. The highest accuracy was achieved at double resolution, with 90.83% for training data and 70% for testing data, compared to triple resolution (90% training, 66.25% testing) and the original resolution (85.83% training, 67.5% testing). The KNN method also outperformed the Naïve Bayes Classifier used in previous research. Therefore, image resolution enhancement and the implementation of the KNN method are proven to be effective in improving the accuracy of corn kernel quality classification.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Biji Jagung, Ekstraksi Fitur Histogram, K-Nearest Neighbor (KNN), Pengolahan Citra Digital
Subjects:	Q Science Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > Q Science (General) > Q337.5 Pattern recognition systems Q Science > QC Physics
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Physics > 45201-(S1) Undergraduate Thesis
Depositing User:	Putri Ayu Kumalahadi
Date Deposited:	17 Jul 2025 07:41
Last Modified:	17 Jul 2025 07:41
URI:	http://repository.its.ac.id/id/eprint/119905

Actions (login required)

View Item