12改进的形态学交替滤波算法-中国图象图形学报.doc

中图分类号：TN957.52 标识码： 文章编号： 论文引用格式： 高分辨率SAR影像形态学层级分析的建筑物检测 张恒1,2,余涛， 1.中国科学院遥感与数字地球研究所，北京00101；.中国科学院大学，北京49 摘 要：：38%、 关键词：建筑物层级分析；形态学滤波；去噪；高分辨率SAR detection of high-resolution SAR image based on morphological hierarchy analysis Zhang Heng1, 2, Yu Tao1, Liu Peng1, Zhang Zhouwei1 1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China 2. University of Chinese Academy of Sciences, Beijing 100049, China Abstract: Objective: The present structure-analysis based algorithms of building detecting in high-resolution SAR images only considers the straight-line and L-shaped buildings, and utilizes the shadowed areas of the highlighted lines on buildings to detect them; in complicated scenes, the shadowed areas cannot be accurately detected by restriction of the dark backgrounds and dense buildings, which results in big errors and low accuracies in building detecting. Focusing on the above problems, an algorithm based on morphological hierarchical analysis of unsupervised building detecting in high-resolution SAR images is developed. Method: The method applies to single polarization high-resolution SAR images. First, the improved alternating sequential filters (Extended ASF) is used to reduce the inherent speckle noise, which eliminate the interference of background noise in homogeneous regions greatly. And then, differential morphological attribute profiles (DAPs) is calculated as the implement of the geometric structure feature extraction of buildings in SAR image under the complex scene. Finally, the post-processing methods like feature fusion and threshold segmentation are performed to get the intricate building information. Result: Compared with other structure-analysis based algorithms, the proposed method maintains a higher detection rate and a lower error for urban area in high density buildings. The precision rate and recall rate of the proposed method reach to 95.38% and 86.31% respectively. Further