[关键词]
[摘要]
提出了一种改进的时间序列极化合成孔径雷达(SAR)图像变化检测方法。该方法首先使用简单线性迭代聚类算法对时间序列极化SAR图像进行过分割,得到时间序列图像的超像素一致表达。然后对各个超像素进行时间序列上的任意双时相交叉变化检测,得到包含时间序列图像变化信息的变化检测矩阵。最后根据变化信息对图像在时间序列上进行平滑滤波并生成动态变化图,对地物变化情况进行分析。实验结果表明,文中方法能够有效地检测和分析时间序列极化SAR图像的变化情况。
[Key word]
[Abstract]
An object-based change detection method for time series polarimetric synthetic aperture radar (PolSAR) images is presented in this paper. First, the improved simple linear iterative clustering (SLIC) algorithm is used to over-segment time series Pol-SAR images for consistent superpixel representation. Then bi-date cross change detections are conducted on all superpixels with statistical likelihood ratio estimation to obtain change detection matrices. Finally, multi-temporal filtering is used to smooth images in time series and the map of dynamic changes is produced by analyzing the change detection matrices. Experiment results show that the proposed method is effective in detecting and analyzing changes in time series polarimetric SAR images.
[中图分类号]
TN957. 52
[基金项目]
国家自然科学基金资助项目(61771351,61331016)
