Dentawyanjana Character Segmentation Using K-Means Clustering CLAHE Adaptive Thresholding Based

Each country has its own historical heritage, one of which is letters/characters. Indonesia has 12 regional characters, and this must be preserved. One of them is the Dentawyanjana character or usually called the Javanese character/carakan. This study aims to segment the Dentawyanjana character using K-means clustering CLAHE adaptive thresholding based. This study consisted of 3 experiments, and each experiment implemented a different hybrid method. The first method is preprocessing which is converting the RGB image to grayscale, then converting it from grayscale to a binary image. After this process, segmentation is carried out using K-means and moment methods. The second method is preprocessed using a grayscale process, then converting it from grayscale to a binary image using adaptive. The results of the adaptive threshold process are clustered using K-means and segmentation is done using the moment method. The third method used begins with standardizing the dimensions to
$800\times 1200$
pixels which are then preprocessed as in the first and second methods converting the RGB image to grayscale. Then the enhancement is done using CLAHE, and the result of this process is continued to be implemented with an adaptive threshold. The segmentation process is clustered using K-means and moment methods. The results show that the accuracy of the first, second, and third methods are 68.23%, 94.18%, and 95.52%. While the resulting AUC values are 0.798, 0.962, and 0.975.