AR谱结构特征提取及其对车辆目标的识别效果分析

针对车辆噪声识别问题，在对实测的４种车辆噪声进行预处理和ＡＲ谱估计的基础上，提取了谱重心，谱面积和卓越频率成分等几个特征量。实测的车辆噪声数据检验表明，基于ＡＲ谱结构提取的这几个特征具有良好的稳定性，且对车辆目标具有良好的识别效果，这对最终完成战场声目标模糊模式识别具有重要的参考价值。

Abstraction of AR Spectrum Structure Features and Analysis on their Effect on Vehicle Target Identification

Aiming at vehicle noise recognition, this paper abstracts some features, such as, spectrum gravity center, spectrum area and prominent frequency ingredient on the basis of making pre processing and AR spectrum estimation of tested four kinds of vehicle noises. Tested data of vehicle noises have shown that the features abstrated on the basis of AR spectrum estimation have good stability and good recognition effectiveness to a vehicle target, which has important value for finishing battle field vocal target fuzzy pattern recognition.