Image Segmentation Research Based on Graph Optimal Binary Tree
Zhou Yua · Jiguang zazhi · 2014
In order to improve the quality of image segmentation, graph optimal binary tree is proposed. Firstly,pixel corresponding graph edge weight is established between nodes, minimum spanning tree edge weight is determined optimal segmentation threshold, length which is larger threshold edge is cut; Secondly, similar relation function of two regions is determined optimal segmentation region. Thirdly, segmentation of region is judged merge; lastly, process is presented. Simulation result shows this algorithm effect segmentation of image is good, isn't affected by external factors, uniform measure is largest.