论文部分内容阅读
水体是合成孔径雷达(SAR)图像解译的一类重要内容。针对含水体的SAR图像的成像特点,给出了一种基于轮廓的配准方法。首先,提出了融合观测图像局部统计信息的自适应权马尔科夫随机场(MRF)分割模型,以分割SAR图像水体目标并提取其精确轮廓。然后,提出了轮廓匹配的非均匀高斯混合模型(GMM),该模型能融合轮廓上点的位置信息和以轮廓点为中心的窗口的灰度相似性信息。最后,对含水体目标的SAR图像进行配准实验。结果显示所提出的MRF分割模型能精确地定位目标边缘并保持图像的细节,轮廓匹配的非均匀GMM对噪声、外点及局部变形具有稳健性,能较好地实现含水体目标的SAR图像配准。
Water body is an important part of synthetic aperture radar (SAR) image interpretation. Aiming at the imaging characteristics of SAR images containing water bodies, a contour-based registration method is presented. Firstly, an adaptive Markov-Markov random field (MRF) segmentation model is proposed to segment the local statistical information of the observed image to segment the target and extract the accurate contour of the SAR image. Then, a contour matching non-uniform Gaussian mixture model (GMM) is proposed, which can fuse the position information of contour points and the gray similarity information of window centered on contour points. Finally, the registration experiments of SAR images containing water bodies were performed. The results show that the proposed MRF segmentation model can accurately locate the target edge and maintain the details of the image. The contour matching non-uniform GMM is robust to noise, outliers and local deformation, and can well realize the SAR image with water body quasi.