基于基因结构树模型的红外图像分割研究

针对红外图像分割的特点,采用基因结构树模型。首先通过由根节点至节点顺序累加获取属性为结构树模型的生长代价;接着判决规则为结构树分支节点距离计算,借助分支节点之间的夹角表示,夹角越小,则分支节点距离越小;最后从基因树中选择其中似然值最大的作为最优树,给出了算法流程。实验仿真结果显示算法对红外图像分割很好地保留目标区域,有较为简洁平滑的边缘,分割质量评价的定量结果优。

Infrared image segmentation method based on genetic structure tree model

according to the characteristics of infrared image segmentation, using the gene structure tree model. First through the growth of the root node to node order accumulation get properties for the structure tree model; then the decision rules for the calculation of the distance tree branch node, the angle between the said branch node, the smaller the angle, the branch node distance is small; the last one the largest likelihood value as the tree from the gene tree, the algorithm flow is given. Simulation results show this algorithm for infrared image segmentation well reserve the target area, there are more concise and smooth edge, quantitative segmentation quality evaluation results and.