We present a color image segmentation algorithm, RCRM, based on the detection of Representative Colors and on Region Merging. The 3D color histogram of the RGB input image is built. Colors are processed in decreasing frequency order and a grouping process is accomplished to gather in the same cluster all colors that are close enough to the current color. Colormapping is done to originate a preliminary image segmentation. Segmentation regions having small size undergo a merging process. Merging is actually accomplished only for adjacent regions whose colors do not significantly differ. The parameters involved by the algorithm are set automatically by taking into account color distribution in the input image and geometrical features of the regions into which the image is partitioned. The algorithm has been tested on a large number of RGB color images originating satisfactory results. © 2013 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Ramella, G., & Sanniti Di Baja, G. (2013). Image segmentation based on representative colors detection and region merging. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7914 LNCS, pp. 175–184). https://doi.org/10.1007/978-3-642-38989-4_18
Mendeley helps you to discover research relevant for your work.