Pengenalan Huruf Hiragana Menggunakan Metode Histogram of Oriented Gradients (HOG) dan Support Vector Machine (SVM)

Mustofa, Tsabita Safana (2025) Pengenalan Huruf Hiragana Menggunakan Metode Histogram of Oriented Gradients (HOG) dan Support Vector Machine (SVM). Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21081010282-cover.pdf Download (1MB)
	Text (Bab 1) 21081010282-bab1.pdf Download (260kB)
	Text (Bab 2) 21081010282-bab2.pdf Restricted to Repository staff only until 19 June 2027. Download (724kB)
	Text (Bab 3) 21081010282-bab3.pdf Restricted to Repository staff only until 19 June 2027. Download (1MB)
	Text (Bab 4) 21081010282-bab4.pdf Restricted to Repository staff only until 19 June 2027. Download (4MB)
	Text (Bab 5) 21081010282-bab5.pdf Download (249kB)
	Text (Daftar Pustaka) 21081010282-daftarpustaka.pdf Download (198kB)

Abstract

Hiragana is the most commonly used basic writing system for writing native Japanese words. Hiragana characters pose unique challenges in the recognition process due to their distinctive shapes. This study aims to develop an automatic Hiragana character recognition system using the HOG method for feature extraction and SVM as the classification method. The dataset used consists of 4,600 Hiragana character images, with 100 images for each character class. During the preprocessing stage, all images are first converted to grayscale format and then resized to 64x64 pixels. Feature extraction was performed using HOG with parameters of 8x8 pixel cell size, 2x2 cell block, and 9 orientation bins. The SVM model was trained using three different kernels with testing of several combinations of C and gamma parameters with varying data proportion divisions to obtain the best performance. The test results show that the model achieves an accuracy of 97.50%, precision of 97.59%, recall of 97.50%, and an F1-score of 97.49% on the HOG-SVM model with an RBF kernel, C = 10, and gamma = 0.01 on an 80:20 dataset split. This performance demonstrates that the HOG and SVM methods are highly effective in recognizing Hiragana characters, even when the characters have similar shapes. The best model was then implemented in a web interface using Flask to facilitate users in trying interactive character recognition. This research is expected to serve as a foundation for the development of Japanese language learning systems based on technology.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Sari, Anggraini Puspita	NIDN0716088605	anggraini.puspita.if@upnjatim.ac.id
Thesis advisor	Maulana, Hendra	NIDN1423128301	hendra.maulana.if@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.6 Computer Programming

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Tsabita Safana Mustofa

Date Deposited:

20 Jun 2025 01:13

Last Modified:

20 Jun 2025 01:13

URI:

https://repository.upnjatim.ac.id/id/eprint/38629