最小模糊熵算法在SAR影像溢油检测中的应用

Application of Minimum Fuzzy Entropy Segmentation Method to Oil Spill Detection in SAR Imagery

针对溢油SAR影像的油膜识别不准确的问题,该文引入最小模糊熵的图像分割方法,构建基于一维灰度直方图与二维灰度直方图的最小模糊熵隶属度函数,并基于模糊熵分割算法对试验影像进行处理。实验结果表明:最小模糊熵的分割算法能够对SAR影像实现比较好的油膜分割效果,且二维的模糊熵分割效果由于加入了图像灰度邻域信息,抗噪能力更强,分割效果要优于一维的模糊熵分割算法。同时,实验验证对于直方图双峰特性不明显的图像,采用最小模糊熵的的分割效果要优于最大模糊熵。

In order to identify the oil film on SAR image,this paper introduced the minimum fuzzy entropy image segmentation method,constructing the membership function based on the 1D-gray level histogram and 2D-gray level histogram respectively,and processed the experiment images based on fuzzy entropy segmentation method.The experimental result shows that the minimum fuzzy entropy image segmentation is more fitting for SAR images,and the 2D segmentation method is better than the 1D segmentation method,because it added the grayscale neighborhood information with strong resistance to noise. Meanwhile,the test shows that the image without two apex characteristic in histogram can get better result using the minimum fuzzy entropy than the maximum fuzzy entropy image segmentation.