分散、递阶蚁群算法及其在相变序列图像分割中的应用

提出了一种分散、递阶蚁群算法，它将多个分散的蚁群并行求解各自对应的子问题，形成一个执行层，而递阶协调层利用执行层反馈的信息协调执行层的优化过程，从而得到递阶求解问题的整体最优解．该算法已应用于相变热图序列图像模糊相变线的提取问题，即运用区域划分方法先将序列图像的分割转化为关联型多子图搜索问题，再利用分散、递阶蚁群算法进行求解，结果表明该算法能很好地利用相变线的运动信息成功地解决热相变序列图像分割问题．与蚁群算法独立搜索比较，分散、递阶蚁群算法能更有效地实现多子图之间的关联．

Decentralized and Hierarchical Ant Colony Optimization and Its Application to Phase Change Image Sequences Segmentation

A class of decentralized and hierarchical ant colony optimization is presented, in which the sub-problem is solved by multiple decentralized ant colonies side by side to form an execute level. The optimization process of the execute level is coordinated in the hierarchical coordination level with the feedback information from the execute level to obtain the general optimal solution of the hierarchical solving problem. The algorithm is applied to the extraction problem of fuzzy phase change lines of phase change thermograph sequences, where the sequences image segment is transformed into search problems of associated multiple sub-images by region partition method, then solved by decentralized and hierarchical ant colony algorithm. The simulation results show that the moving information of phase change lines can be utilized well to solve the segment problem of phase change thermograph by the proposed algorithm. Compared with the single ant colony algorithm, it can more effectively solve the correlation problem among multiple subgraphs.