基于邻域信息的SAR图像变化检测

为了减少合成孔径雷达(SAR)图像中乘性斑点噪声对变化检测结果的影响,充分地利用了像素的邻域信息。首先使用邻域比值(NR)方法构造差异图像,然后提出基于邻域信息的模糊C均值聚类(FCM)算法。NR算子在构造差异图像时能够较好地保留图像信息并抑制噪声的干扰。同时将邻域信息引入到FCM算法的目标函数,以邻域加权距离改进了FCM算法在欧式距离计算中的不足,并约束了隶属度函数,减少了噪声对邻域中心像素的干扰。通过以上考虑像素邻域信息的算法,得到了差异图像的聚类结果,从而实现了SAR图像的变化检测。实验结果表明,所提算法较传统的FCM和K-means聚类算法,可以较好地保留图像变化区域的信息,同时提高了SAR图像变化检测的准确度。

Change detection for SAR images based on neighborhood information

In order to reduce the influence of multiplicative speckle noise on the change detection results in synthetic aperture radar(SAR)images,this paper makes full use of pixel neighborhood information.Firstly,the neighborhood-based ratio(NR)method is used to construct the difference image,and then a fuzzy C-means clustering(FCM)algorithm based on neighborhood information is proposed.The NR method can better retain image information and suppress noise interference when constructing difference images.At the same time,the neighborhood information is introduced into the objective function of FCM algorithm.The neighborhood-weighted distance improves the shortcomings of FCM algorithm in the calculation of Euclidean distance,constrains the membership function,and reduces the interference of noise on the central pixels of the neighborhood.Through the above algorithm considering the pixel neighborhood information,the clustering results of difference images are obtained,and the change detection of SAR image is realized.Experimental results show that,compared with traditional FCM and K-means clustering algorithms,the proposed algorithm can better retain the information of image change area,and improve the accuracy of SAR image change detection.