摘要
提出一种基于随机游走的遥感图像变化检测与配准方法。该方法的特点是,提取多时相遥感图像的关键点,作为随机游走的初值种子点和图像配准的特征点,将图像配准和变化检测结合起来,提高变化检测的效率。该方法采用一种能量最小化策略来提取像素的不透明度,能检测到大范围的变化,包括一些细微的变化。仿真表明了该方法的有效性。
A random walk-based change detection and registration method for remote sensing image is proposed. It has the following characteristics： extracting the key points of multi-temporal remote sensing image as the random walking initial seed nodes and the feature nodes, and combining the image registration with change detection, this improves the efficiency of change detection. An energy minimisation scheme is employed by the method to extract the pixel opacity, and the large scale change is able to be detected including some subtle changes. Simulation proves the validity of the proposed method.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第8期283-286,306,共5页
Computer Applications and Software
关键词
变化检测
图像配准
随机游走
Change detection Image registration Random walk
