一种基于模糊连接度的区域生长毁伤特征提取算法

针对传统的区域生长算法在提取军事目标毁伤时,不能较好地处理毁伤区域与背景之间的模糊性,生长结果对种子点的选取敏感等问题,提出了一种基于模糊连接度的区域生长算法。算法以模糊连接度作为生长点与种子点相似度的度量,有效地保持了像素之间的模糊性,按照模糊连接度由强至弱的顺序生长,使得生长的结果对种子点的选取不敏感;采用灰度波动控制和USAN面积相结合的方式作为生长准则,在适应目标区域灰度变化的同时得到了较好的边缘定位效果。实验结果证明了算法的有效性。

A Characteristic Distill Arithmetic of Region Growing Method Based on Fuzzy Connectedness

To solve the problem that the traditional region growing algorithm can not deal well with the fuzziness between the damage area and the background,and the segmentation result is sensititive to the initial seed pixel,a region growing method based on fuzzy connectedness is proposed. The method takes fuzzy connectedness as the similarity measure between the initial pixel and the growing pixel, which keeps the fuzziness among the pixels well. A growing crition integrated gray- fluctuation control with USAN area is proposed to fit the gray fluctuation in the damage area and get edge position precisely. The experiment results show the validation of this method.