电光与控制, 2017, 24 (4): 5, 网络出版: 2021-01-25
小波包特征能量算子与多核函数组合KPCA的声目标识别
Battlefield Acoustic Target Recognition Based on Wavelet Packet Analysis and Principal Component Analysis
肓音探测 小波包特征能量算子 多核函数组合 核主成分分析 sound detection wavelet packet teager operator multi-kernel function combination kernel principal component analysis
摘要
提出了一种小波包特征能量算子和多核函数组合KPCA的声目标特征参数提取方法。首先对声目标信号采用小波包能量特征算子进行特征参数提取, 然后将组合核函数应用于核主成分分析。实验数据表明, 基于小波包能量特征和多核函数组合KPCA特征参数不仅大大降低了特征向量的维数, 并且有效地提高了识别率, 降低了计算复杂度。
Abstract
A new method was proposed for feature extraction from acoustic targets by combining wavelet packet characteristic energy operator with multi-kernel function combination Kernel Principal Component Analysis (KPCA). First, wavelet packet characteristic energy operator was used to the acoustic signal of target for extracting characteristic parameters. Then, the combined kernel function was used for KPCA. The experiment results show that:The characteristic feature extraction method proposed can reduce the dimensions of the feature parameters, improve recognition rate and decrease computation complexity.
曾番, 黄文龙, 夏伟鹏, 冯卉. 小波包特征能量算子与多核函数组合KPCA的声目标识别[J]. 电光与控制, 2017, 24(4): 5. ZENG Fan, HUANG Wen-long, XIA Wei-peng, FENG Hui. Battlefield Acoustic Target Recognition Based on Wavelet Packet Analysis and Principal Component Analysis[J]. Electronics Optics & Control, 2017, 24(4): 5.