模糊连通度分割算法在探地雷达图像分析中的应用

Application of a New Ground Penetrating Radar Image Analysis Method Based on Fuzzy Connectedness

对基于相对模糊连通度的交互式图像分割方法进行简化,并将改进后的算法应用到探地雷达图像的分割中,提高了地下目标物体的提取速度。此外,还提出一种探地雷达图像的预处理方案,通过小波局部软阈值去噪算法来抑制探地雷达回波中的干扰信号,再利用窗宽/窗位调整算法突出图像中目标的灰度信息,使图像中的目标得以增强。从而得到一种快速有效的地下目标物体提取方法。用实测探地雷达图像所做的测试实验取得了令人满意的结果。

Some simplification and optimization were made in the interactive segmentation algorithm based on relative fuzzy connectedness, thus resulting in a highly efficient segmentation algorithm without lowering the segmentation precision. And this modified segmentation algorithm is applied to Ground Penetrating Radar （GPR） image analysis. In addition, a pre-processing method is suggested, and the objects in GPR images are enhanced through introducing the level-dependent soft-thresholding wavelet de-noising algorithm and slide window processing. Thus the extraction of GPR image from complex background can be achieved. The experimental results show that the proposed algorithm can extract the objects of interest from the actual GPR image quickly and reliably.