EN 
IN
Journal article
Color Image Segmentation using Kohonen Self-Organizing Map (SOM)
I Komang Ariana RUKMI SARI HARTATI I Ketut Gede Darma Putra NI KADEK AYU WIRDIANI
Volume : 6 Nomor : 2 Published : 2014, May
International Journal of Engineering and Technology (IJET)
Abstrak
Color image segmentation using Kohonen Self-Organizing Map (SOM), is proposed in this study. RGB color space is used as input in the process of clustering by SOM. Measurement of the distance between the weight vector and the input vector in the learning and recognition stages in SOM method, uses Normalized Euclidean Distance. Then, the validity of clustering result is tested by Davies-Bouldin Index (DBI) and Validity Measure (VM) to determine the most optimal number of cluster. The clustering result, according to the most optimal number of cluster, is processed with spatial operations. The operations are used to eliminate noise and the small spatial regions which are formed from the clustering result. This system allows the segmentation process become automatic and unsupervised. The segmentation results are close to human perception.