Analisis Segmentasi Warna pada Citra Dua Dimensi Dengan Pendekatan Komputasi Fotografi

Abstract

Color segmentation techniques on two-dimensional image was originally used in digital image processing applications in engineering sciences. Recent years, these techniques have been widely used in other disciplines. Photography in the digital era use this technique in a variety of needs, as an example for selecting objects based on color pixel elements. This research was done with computational photography approach to perform selection of object area based on color elements in each pixel. Observations and analysis done by comparing three common color segmentation method, namely k-means clustering method, index coloring, and Giannakopoulos HSV color detection. This research is a preliminary project of automation process for body segmentation based on colors. After testing, it turns out segmentation based on Giannakopoulos HSV color detection has added value. This is because the ease of data training process to determined the classification of color values and tolerance segmentation based on intensity variation from selected color.