探地雷达目标检测中的加窗统计量定位方法

An Algorithm of Window Choice in Depth by Statistics Feature in GPR Target Detection and Location

目标检测与定位是探地雷达的一个重要问题。目前,多数检测与定位方法都是通过判断探地雷达信号中的双曲线特征实现的。在检测目标大致区域的基础上,要实现目标定位,无论是合成孔径算法还是叠加-幅度波速法,都存在计算量大的问题。现有的统计量方法还需依赖霍夫变换方法检测出双曲线,算法也较复杂。根据探测背景介质结构相似及在探测方向上利用整个A-scan进行统计量计算干扰较大的特点,提出了一种探地雷达目标快速定位方法。首先以能量和方差为统计量,在深度方向上判断出目标区域的时间窗口,而后在探测方向对A-scan加窗,只选择该窗口下的部分A-scan数据,以能量为统计量确定目标的位置。由于仅采用方差、能量两个统计量进行计算,算法实现比较简单,并且采用时间加窗方法去除了回波数据上大部分与目标无关数据的干扰,对目标的定位准确度较高。实测数据表明,该方法适用于探地雷达目标检测中对目标的快速准确定位。

Target detection and location is important in the application of the ground penetrating radar. Target location is based on the result of the detection, which in turn depends on the hyperbolas of the ground penetration radar. Many algorithms for this purpose involve a large amour of computational effort, such as SAR and overlay amplitude-velocity algorithm. They include an algorithm on window- statistics to consider noises and non-targets data. At the same time, Hough transform algorithm has also to be used. Based on the similarity of time samples of ground penetrating radar and the fact that the possible target lies in some area, an algorithm is proposed in this paper for ground penetrating radar target detection and location. First, the algorithm considers energy and variance as statistics features of one-dimensional data in depth, to determine the probable depth of target signals by the right threshold. Then the energy variant is used as the statistics feature of an A-scan in the above special depth window by the threshold. Finally, the target location is determined. Because it only uses a part of A-scan data without considering the non-targets data, the location is more accurately determined than the algorithm that uses the whole A-scan data. Moreover, the algorithm is more effective than above mentioned other algorithms. The result of the test shows that the algorithm is very effective in the fast target detection and location.