快速精确地检测SAR图像的变化是SAR应用亟待解决的问题。针对这一问题,提出了基于区域热核不变量的SAR图像变化检测方法。该方法利用图上热核的特征,可以充分发挥图谱理论的优点,且计算简便、矩阵的扰动性小。通过模拟分析研究了热核的特征,并应用于实际SAR图像的变化检测。实验结果表明该方法可以快速精确地检测到变化区域,且其变化检测的精度要优于区域似然比检测方法。
The fast and accurate detection of synthetic aperture radar (SAR) image change has been a pressing problem to SAR image applications. In this paper, a change detection algorithm based on region heat kernel invariants is proposed. This method used to study the characteristics of heat kernel on graphs can allow full play to advantages of graph spectral theory. It is convenient to compute and has small matrix perturbation. The characteristics of heat kernel are studied through the simulation images and applied in SAR images change detection. Experiment results show that the change areas can be detected fast and accurately by our method. Meanwhile, comparing with the region likelihood ratio method, our method can get a better accuracy of change detection.