一种稳健的雷达高分辨距离像目标识别算法

雷达高分辨距离像目标识别算法通常对目标回波的噪声大小比较敏感，如果测试样本和训练样本的信噪比不等，那么将会导致识别性能的下降。在实际应用中，需要识别算法在不同噪声强度下都能够保持稳健的性能，因此在概率主分量分析模型的基础上，提出一种稳健的雷达高分辨距离像自动目标识别算法。该算法能够让模型随着噪声强度的不同而自适应地调整其参数，并且分析了雷达数据的能量归一化处理对模型参数的影响。由于算法搜索时间较长，为提高算法的搜索效率，推导了一个快速算法。基于实测数据的仿真实验结果验证了方法的有效性, 对噪声有较好的稳健性。

Robust algorithm for radar HRRP target recognition

The algorithms of radar high resolution range profile (HRRP) are often sensitively to noise, the performance of recognition will descend if the signal to noise ratio of the testing sample is not equal to as the training sample. In practical application, the algorithms should have a robust recognition performance in different noisy conditions. Thus, based on the probabilistic principal component analysis (PPCA) model, a roust algorithm for radar target recognition is proposed. This algorithm makes the parameters of PPCA model alter by different noisy conditions, and the influence of normalization of radar data on the parameters of PPCA model is analyzed to improve the searching efficiency of the algorithm, a fast algorithm is deduced. Experimental results for measured data show that the validity of the proposed method, which is more robust to noise.