一种基于区域生长的目标提取算法

An Object Extraction Method Based on Region Growing

对利用区域生长进行目标提取问题进行了研究,提出一种基于模糊连接度的生长算法,该算法以模糊连接度作为生长的原则,采用一种自适应阈值与USAN面积相结合的方式作为生长终止规则,通过具体的实验,证明该算法的提取结果对种子点的选取不敏感,算法对区域的边缘具有较好的定位能力,有效地防止了过分割的情况.

This paper researches on the region growing methods, It presents a novel region growing methods based on fuzzy connectedness. This method takes the fuzzy connectedness as the region growing principle and the combination of an adaptive threshold and usan area as the terminal principle. Through specific experiments, it is proved that the growing result is not sensitive to the seed points and the method prevents over segmentation effectively.