中文摘要：

车载前视超宽带地表穿透雷达在地雷探测中遇到的困难是虚警率过高,地雷与杂波在全孔径图像中很难准确区分。为降低地雷探测过程中的虚警率,该文提出一种基于目标子孔径图像方位不变性的检测方法。该方法利用分裂发射虚拟孔径成像模型,将全孔径图像分解为左右两个子孔径图像,并根据子孔径图像中的目标一维距离剖线建立双峰特征模型。在此模型基础上提取具有方位不变性的若干特征,进而得到左右子孔径图像中目标的一致性度量,并将该一致性度量作为最终的特征向量送入鉴别器加以判别。经实测数据验证,该算法能有效剔除原先在全孔径图像中无法剔除的杂波,从而降低前视地表穿透合成孔径雷达中地雷探测的虚警率。

英文摘要：

The difficulty of landmine detection using vehicle-mounted ultra-wideband ground penetrating radar was the excessively high false alarm rate.It was hard to distinguish landmine and clutter in full aperture.To decrease the false alarm rate during landmine detection,a detection algorithm based on aspect invariant characteristics of sub-aperture images was proposed in this paper.With the split transmitting virtual aperture imaging model,the full image was decomposed into left and right sub-aperture images.And the double-hump model was established according to the one dimension range profile of sub-aperture images.Based on this model,some aspect invariant characteristics features could be extracted.Furthermore,the measurement of consistency of left and right sub- aperture images was obtained and sent to classifier as the final feature vector.It was proved by real data that the algorithm in this paper can effectively eliminate the clutter which can not be eliminated in the full aperture image.So this algorithm can decrease the false alarm rate in vehicle-mounted ultra-wideband ground penetrating radar.