Hierarchical Region-Based Segmentation by Ratio Contour

Ratio Contour algorithm is recently introduced into the research about image segmentation. Several papers show that Ratio Contour has good performance on salient boundary detection, which can reach global optimization. This paper aims to solve hierarchical region-based segmentation by Ratio Contour algorithm. Two important problems should be addressed for this purpose: One is extending boundary-based segmentation to region-based segmentation. The other is solving open boundary detection by Ratio Contour. For the first problem, we implement it by using graph partition technology based on boundary detection results. About the second problem, we adjust the constructed graph by adding a special vertex, boundary-vertex. Every detected boundary is looked as a kind of special vertex in the graph. The experiments on synthesized and real images show that the presented method has good performance.