Image Segmentation by Grouping Pixels in Color and Image Space Simultaneously using DBSCAN Clustering Algorithm

Resumen

In this paper, the authors propose a technique for image segmentation by integrating the spatial connectivity and color features of pixels. Taking into account that an image can be considered as a dataset in which each pixel has a spatial location and a color value, color image segmentation can be obtained by clustering these pixels into different groups of coherent spatial connectivity and color. To discover clusters in spatial databases in data mining, density-based clustering is used. Here, the authors have used density-based clustering (DBSCAN) to determine the spatial connectivity of the pixels. Color similarity of the pixels is measured in perceptually uniform Munsell (HVC) color space under NBS color distance measurements. Experimental results using the proposed method show encouraging performance. Keywords: image segmentation, DBSCAN, NBS color-distance, color difference, HVC color space