多尺度分割优化的SAR变化检测
SAR Change Detection by Multi-scale Segmentation and Optimization
-
摘要: 基于SAR图像的变化检测和尺度有较强的相关性,提出了一种联合变化前后的两幅SAR图像进行多尺度分割的方法,在分割的图斑基础上采用了交叉熵计算图斑之间的差异程度。根据多个尺度分割的斑块计算系列差异指数,采用多尺度优化的方法得到最佳差异指数,最终实现变化区域的提取。通过对SAR图像的实验处理,说明了在多尺度联合分割的基础上进行变化检测可以得到较好的检测效果。Abstract: A multi-scale segmentation method based on two temporal SAR images is proposed.The segmented parcels are used to calculate divergence index by Kullback-Leibler divergence and optimal divergence index is obtained according to several divergence indexes calculated using parcel in different scale.Change areas are detected based on the optimal divergence index image.The processing test demonstrates that the detected result using this method is satisfactory.