SAR与TM图像自动配准算法研究

Automatic Registration of SAR and TM Images

为了克服SAR(合成孔径雷达)和TM(专题制图仪)图像配准时受SAR图像中相干斑噪声影响大,以及手工选取配准点精度低的缺点,提出了一种基于闭合区域的图像配准方法,先对SAR图像和TM图像进行处理得到闭合边缘图像后,采用区域生长的方法确定闭合区域,并以其所含像素点数为依据对闭合区域进行匹配,实现图像配准。实验结果表明:该方法稳定有效,对于待配准图像的波段无限制,适应性强。

Aiming at the fact that the registration of Synthetic Aperture Radar （SAR） and thematic mapper（TM） image is disturbed by the speckle noise in SAR images, and the low accuracy of manual selection of registration points, this paper proposes a new image registration method based on the closed region. This method includes three steps： first, processing SAR image and TM image to get their edge image whose region are closed; second, defining the closed region with region-growing algorithm; at last, matching the closed region on the basis of the number of their contained pixel to realize their registration. The results of experiment show that the method presented is not only stable and efficient, but also without limitation on the waveband of the images under registration, and has strong flexibility.