项目名称: 面向图像网状结构体的蚁群分割算法
项目编号: No.61773297
项目类型: 面上项目
立项/批准年度: 2018
项目学科: 其他
项目作者: 陈黎
作者单位: 武汉科技大学
项目金额: 16万元
中文摘要: 本项目以医学图像、遥感图像、显微图像等领域中存在的网状结构体为新型研究对象。针对现有图像分割算法通用性不强、容易造成过分割或欠分割的局限性,以蚁群算法为核心,发展线型运动演化模型和个体智能学习两个特长,提出能充分利用网状体连接拓扑特性进行图像分割的蚁群分割算法。具体研究目标是针对网状结构体的基本结构是狭长线条这一特点,提出研究蚂蚁沿网状体移动的线型运动模型,结合运动惯性约束研究蚂蚁运动在采样图上的演化机制。
中文关键词: 结构描述与匹配;蚁群;图像分割;网状结构体;血管网络
英文摘要: The objective of this project is to study segmenting network-structure object in image, in areas of medical imaging, remote sensing and microscopic imaging. Conventional image segmentation approaches usually produce over-segmentation or under-segmentation results. This project exploits the ant colony optimization algorithm, with the proposed linear transition evolution model and individual intelligent learning, to exploit the topology features of network-structure object to perform image segmentation. This project studies the following fundamental issues. To exploit the narrow line feature of the network-structure object, a new linear transition model is proposed to guide the movement of ant colony. Furthermore, it is incorporated with the conventional Markov chain algorithm with inertia constraint to study the evolution mechanism of ant movement in the image.
英文关键词: Structure descripition and matching;Ant colony;Image segmentation;Network-structure;Vascular network