结合KI准则与数学形态学滤波的SAR变化检测

Change detection of SAR images based on KI criterion and mathematical morphology filter

合成孔径雷达(Synthetic Aperture Radar,SAR)影像因其独特的成像方式,而具有明显的斑点噪声,这限制了SAR影像变化检测的精度和应用。在广义高斯模型假设下,结合KI(Kittler-Illingworth,KI)最小错误率准则计算自动阈值,得到SAR影像变化检测图,利用二值数学形态学滤波算法对变化检测图进行后处理,并分析了方形和圆形结构元素在不同结构尺寸下的滤波效果。在两组SAR数据上的实验表明,选用恰当的结构元素与大小,可以有效提高变化检测精度。

Synthetic aperture radar （SAR） image had obvious speckle noise because of its unique imaging mode, which limited the accuracy and application of SAR change detection. Under the assumption of the generalized Gaussian model, combined with the Kittler - Illingworth minimum error rate criterion automatically calculated threshold to get SAR image change detection map. And then used a binary mathematical morphology filtering algorithm for the change detection map, and analyzed the filter effects of square and disk structural elements in different sizes. The experiment on the SAR data of the two groups shows that the appropriate structural elements and sizes can be used to improve the accuracy of the change detection effectively.