Color Image Segmentation Using Fast Density-Based Clustering Method

摘要

Color image segmentation is an important research topic in the field of computer vision. In this paper, we propose a method for image segmentation by computing similarity coefficient in RGB color space. Then, we apply the density-based clustering algorithm TI-DBSCAN on regions growing rules that in turn speeds up the process. This new method has three advantages. First, this method can reduce the disturbance of noise and get the segmentation numbers more accurately. Second, it needn't to change the RGB color space to other space. Third, it uses a triangle inequality property to quickly reduce the neighborhood search space. The experimental results illustrate that the new approach segmentation method can efficiently segment image.