In this paper, image segmentation is based on hybrid correlation clustering. Correlation-clustering is a graph-partitioning algorithm used in natural language processing, document clustering and image segmentation. In this proposed method the hybrid correlation clustering improves the performance and accuracy of the existing higher-order correlation clustering. First apply the higher-order correlation clustering over Hypergraph, S-SVM, and combine the difference of existing and proposed algorithms. Experimental results on SBD dataset shows that the proposed method allow to archieve state-of-the-art results with a simpler and efficient model than the previous work.