Implementasi Glcm (Gray Level Co-Occurrence Matrix) Dan Algoritma K-Nearest Neighbors Dalam Klasifikasi Kualitas Hasil Pengeringan Cengkeh

SERA, Maria Yeve Desri (2023) Implementasi Glcm (Gray Level Co-Occurrence Matrix) Dan Algoritma K-Nearest Neighbors Dalam Klasifikasi Kualitas Hasil Pengeringan Cengkeh. Undergraduate thesis, Universitas Katolik Widya Mandira Kupang.

	Text ABSTRAK.pdf Download (447kB)
	Text Bab I.pdf Download (227kB)
	Text Bab II.pdf Restricted to Repository staff only Download (683kB)
	Text Bab III.pdf Restricted to Repository staff only Download (661kB)
	Text Bab IV.pdf Restricted to Repository staff only Download (916kB)
	Text Bab V.pdf Restricted to Repository staff only Download (748kB)
	Text Bab VI.pdf Download (1MB)

Abstract

Clove is one type of spice plant that has many benefits, including being used as a raw material for the pharmaceutical industry, food industry and as a raw material for cigarette mixtures and so on. But before the cloves are processed, it can be seen from the good and bad quality of the cloves. The purpose of this study was to identify the quality of good cloves and bad cloves by text feature extraction GLCM (Gray Level Co-Occurrence Matrix) and K-Nearest Neighbors (KNN). The image sample data used in this study amounted to 1600 images consisting of good clove and bad clove images, which will then be divided into training data and test data. The training data sample is 1000 images divided by 500 good clove images and 500 bad clove images, while the test data sample is 600 images divided by 300 good clove images and 300 bad clove images, for sample data the image collection consists of 20 images consisting of 10 sets of dominant images of good cloves and 10 sets of dominant images of bad cloves will be tested for this data. Texture feature analysis was performed on the skin using GLCM for the extraction and training process to obtain results. The results of this study indicate that the use of GLCM (Gray Level Co-Occurrence Matrix) texture feature extraction with Contrast, Correlation, Energy and Homogeneity features was successfully applied with an accuracy value of 91.50% for training, 85.50% for testing, and for group testing image by 95%.

Actions (login required)

View Item