主分量分析子空间模型和概率主分量分析模型已成功应用于雷达高分辨距离像(HRRP)统计目标识别中。传统的统计建模分帧方法都是等间隔角域分帧建模。提出了一种基于子空间模型分帧方法,首先将距离像分成若干子帧,然后将子空间夹角与均值欧式距离的融合作为度量准则,依次合并度量值最小的相邻子帧来完成分帧。基于实测数据的实验表明该方法得到的分帧结果数量比等间隔分帧的数量少,识别过程中计算量减小,同时识别率也有明显提高。
Principle component analysis and probabilistic principle component analysis models have been successfully used for radar HRRP target recognition. Traditional statistical modeling approaches segmented continuous radar HRRPs into several frames by equal interval partition method. A frame segmentation meth- od based on subspace model is presented in this paper. Firstly, radar HRRPs are segmented into some sub frames. Then the subspace angle and mean Euclidian distance between adjacent subframes are computed as a merging measurement. Finally, the adjacent subframes which have the minimum measurement are merged it eratively to finish the frame segmentation. The experiments based on the measured data obtain fewer frame numbers and higher recognition accuracy.