基于小波能量特征的探雷方法研究

Mine Detection and Identification Using Wavelet Analysis

根据对探地雷达回波信号的分析 ,提出了利用小波域能量特征进行地雷探测的新方法。整个过程包括预处理、特征提取及目标分类三个步骤。在预处理阶段 ,使用波头相关法进行延时校正 ,设计了维纳滤波器滤除系统及环境噪声 ,并采用瞬时频率估计选取含有目标的信号 ;在特征提取阶段 ,采用小波包分解方法并按频段的能量和形成目标的特征提取 ;在识别阶段 ,利用 Fisher线性判别函数和 Mahalanobis距离对目标进行分类研究。将地雷目标及大小与其相近的物体对比试验的数据进行分析处理 ,结果表明 ,提取的特征能有效地将地雷目标与干扰物区分开 ,提高了地雷目标的探测率。

Based on the analysis on the echo from buried objects, a novel method for locating buried mines using pattern recognition is presented. The process comprises pre-processing stage, feature-extraction stage, and classification stage. In the pre-processing stage, wiener filter is designed to reduce noise and to align the signal correlation, and instant frequency is employed to select the signal including objects. In character extraction, the wavelet package energy character vector is extracted. In classification stage, Fisher linear criterion function and Mahalanobis distance are used to study the extracted feature′s separability. The method is tested on data containing mines and some other objects which are like mines in shape and size, and results can greatly improve mine detection.