| 基于多尺度特征的SAR区域型目标检测方法 |
| |
| 引用本文： | 陆立明,王润生,李武皋. 基于多尺度特征的SAR区域型目标检测方法[J]. 雷达科学与技术, 2004, 2(6): 355-359 |
| |
| 作者姓名： | 陆立明 王润生 李武皋 |
| |
| 作者单位： | 国防科技大学ATR重点实验室,湖南,长沙,410073;江南遥感应用研究所,上海,200072 |
| |
| 摘 要： | 通过对SAR图像的目标特性分析，提出了一种SAR区域型目标的检测方法。与通常的一些目标检测算法相比，该算法在ROI的提取和目标建模两个方面作了改进，提出了一个新的特征提取算子并引入多尺度的梯度特征作为分类依据。算法应用于真实的SAR图像进行居民区的检测，获得了预期的结果。实验表明，基于多尺度梯度特征的目标检测算法具有良好的性能。
|
| 关 键 词： | 合成孔径雷达 自动目标识别 方向梯度 目标检测 |
| 文章编号： | 1672-2337(2004)06-0355-05 |
| 修稿时间： | 2004-06-30 |
SAR Regional Object Detection Based on Multi-Scale Features |
| |
| LU Li-ming ,WANG Run-sheng ,LI Wu-gao. SAR Regional Object Detection Based on Multi-Scale Features[J]. Radar Science and Technology, 2004, 2(6): 355-359 |
| |
| Authors: | LU Li-ming WANG Run-sheng LI Wu-gao |
| |
| Affiliation: | LU Li-ming 1,WANG Run-sheng 1,LI Wu-gao 2 |
| |
| Abstract: | A SAR regional object detection approach is presented in this paper. There are two key issues in it:extraction of the regions of interest(ROI) and object modeling. In extraction of the ROI,a new edge detector and a grouping segmentation method based on multi-scale gradient features are developed. In object modeling, semantic knowledge is integrated as the object model. The new approach is applied to real SAR images, and the expected results are obtained. It shows that the approach based on multi-scale gradient features has better performance, and regional objects such as residential areas can be detected. |
| |
| Keywords: | synthetic aperture radar(SAR) automatic target recognition(ATR) direction gradient object detection |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《雷达科学与技术》浏览原始摘要信息 |
|
点击此处可从《雷达科学与技术》下载全文 |
