欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对合成孔径雷达（Synthetic Aperture Radar,简称SAR）图像提出了一种基于小波域改进最小割（Im-proved Minimum Cut,简称IMC）模型的多尺度配准方法.该方法首先将小波分解后的高频信息与SAR图像的特性相结合,用以指导在低频子图像中基于IMC模型的图像分割,从而获得两幅图像中完整的区域轮廓特征,然后通过融合开闭轮廓的信息,分别利用统计直方图和Hausdorff距离获取配准参数并对之细化,最后通过逐层校正参数完成多尺度的配准.实验结果表明,该方法能避免斑点噪声对轮廓提取和匹配的影响,能够实现具有平移、尺度和旋转变化的亚像素级精度的SAR图像配准.
英文摘要：
A multiscale registration method based on improved minimum cut（IMC） model in wavelet domain is proposed for synthetic aperture radar（SAR） images.First,combining the high-frequency information after wavelet decomposition with the characteristic of SAR,an image segmentation method based on IMC model in the low-frequency sub-image is presented.Then,integrated region contours are extracted from reference image and sensed image respectively.Considering the information of open and closed contours,registration parameters are obtained and refined using histogram and Hausdorff distance,respectively.Finally,multiscale registration is realized through revising the parameters in each layer.Experimental results indicate that this method can avoid the influence caused by speckle in contours extraction and matching,also it can realize sub-pixel registration for images differ by translation,rotation and scaling.
同期刊论文项目
同项目期刊论文
位置：