| 摘要: |
| 针对穿墙雷达接收到的回波信号较弱，成像中的耦合信号有所增强，传统动目标检测算法不能完全消除检测后图像中虚警现象的问题，提出一种改进的动态阈值ViBe检测算法，使其能适应穿墙雷达图像动目标像素值变化问题；采用三帧差，消除由墙体和天线耦合等强杂波导致的“鬼影”现象，并利用目标的扩展特性，采取保留连通域的方法，消除了图像中的孤立噪点。实验结果表明，该方法可以有效消除室内强杂波引起的虚警现象，准确、实时地提供目标的位置信息，并具有一定的鲁棒性。 |
| 关键词: 穿墙雷达 动目标检测 视觉背景提取 帧间差分 |
| DOI：DOI:10.3969/j.issn.1672-2337.2019.04.013 |
| 分类号:TN957.52 |
| 基金项目:国家自然科学基金(No.61461012)；广西自然科学基金(No.2016GXNSFAA380036)；广西无线宽带通信与信号处理重点实验室主任基金项目(No.GXKL06160110)；广西科技开发项目(No.1599005-2-1) |
|
| Dynamic Threshold ViBe Algorithm for Moving Target Detection of Through-Wall Radar |
|
ZHENG Zhipeng，LIU Qinghua，JIN Liangnian
|
|
School of Information and Communication，Guilin University of Electronic Technology，Guilin 541004，China
|
| Abstract: |
| For the problems such as weak echo signals received by the through-wall radar，enhanced coupled signals in the image，and the false alarm phenomenon in the detected image that the traditional moving target detection algorithm cannot completely eliminate，we propose an improved dynamic threshold ViBe detection algorithm. It can be adapted to the problem of the pixel value change of the moving target in through-wall radar image. The three-frame difference is used to eliminate the “ghosting” phenomenon caused by strong clutter such as wall and antenna coupling. The isolated noises in the image are eliminated by taking advantage of the extended characteristics of the target and by means of reserving the connecting domains. The experimental results show that the proposed method can effectively eliminate the false alarm caused by strong clutter in the room，provide accurate and real-time location information of the target，and have certain robustness. |
| Key words: through-wall radar moving target detection visual background extractor inter-frame difference entropy |
