IMAGE CLUSTERING BASED ON DISTRIBUTION FUNCTION Husty Serviana Husain, Sapto Wahyu Indratno
Institut Teknologi Bandung
Abstract
Image clustering has been widely studied for a long time and has resulted in many methods. The k-means algorithm is known to cluster extensive data efficiently but is limited to numerical data types. In this study, a new approach was carried out, the procedure for quantifying image data, so it is possible to process it using the k-means algorithm. It is known that the image is a collection of Red, Green Blue pixels with an intensity value of 0-255. The method to perform image clustering does not use numeric data but distribution functions to calculate the similarity of images to one another. The proposed approach represents the image as a probability density function of the pixel intensity value.
Keywords: k-means, image, probability density function
