Color Image Segmentation Using Voronoi Diagram and 2D Histogram

Segmentation divides an image into distinct regions where pixels in each region are characterised by characterised by the same feature. Even though the domain faces various challenges, this work tends to present an improved segmentation method for color image. The proposed algorithm is based on generalized Voronoi diagrams (VD), a renowned technique in computational geometry, which creates clusters of intensity values using information from the vertices of the external boundary of Delaunay Triangulations (DT). The outer boundary of Delaunay triangulation is the Convex Hull of the dotted clouds of the 2D histogram. Unlike the other works, Voronoi diagram is applied on 2D binary image histograms. Therefore, we overcome the fact that the classical 1D histogram ignores the spatial relationship information of the pixels that can give incorrect results. We have obtained promising results in addition to a fast time execution.