【摘要】 为提高合成孔径雷达目标识别性能,提出一种基于目标轮廓匹配的目标识别方法,目标轮廓描述目标的几何外形分布以及尺寸信息,是广泛应用于目标识别的特征之一。将目标轮廓进行分段,采用部分Hausdorff距离度量各个分段的距离,通过科学综合各个分段的部分Hausdorff距离,设计目标轮廓的距离测度,根据具有最小距离的原则判断目标类型。分段匹配的策略有效避免了轮廓提取误差以及目标部分遮挡对于目标识别的不良影响,提高了识别算法的稳健性。通过在MSTAR数据集的多种条件下进行实验,验证了提出方法有效性和稳健性。更多还原

【Abstract】 To improve the performance of synthetic aperture radar(SAR)target recognition,a target recognition method based on matching of target contour was proposed.Target contour reflects the physical sizes and geometrical shape of the target,thus providing reliable descriptions for target recognition.The whole target contour was segmented into several parts and the partial Hausdorff distance(PHD)was used as the distance measure for each part.By reasonably combining individual distances,a distance measure was designed for target contours and the class with the minimum distance was determined to be the target type of the test sample.The strategy of segmenting the whole target contour to several parts can avoid the negative influences brought by contour extraction errors and possible occlusion,thus improving the robustness of SAR target recognition method.Experimental results on the moving and stationary target acquisition and recognition(MSTAR)dataset under several conditions show the effectiveness and robustness of the propose method.更多还原