摘要
为从背景杂波中有效地提取出目标的特征信号,文中提出了一种将小波包分解和神经网络相结合的去噪方案。利用小波包对信号的低频和高频部分进行精细分解,得到信号在多尺度空间上的分解系数。根据信号和噪声在不同尺度小波包分解下的小波包特性,利用神经网络对分解系数进行处理,再用新的小波包分解系数对信号进行重构,以达到滤除噪声的目的。
In order to detect the characteristic signals of targets buried in noises, this paper bring forward a new way to remove the noises, which was based on the theories of wavelet packet transform and neural network. The wavelet packet transform has excellent time-frequency localization analysis capability. It analyses the frequency spectrum of signals and noises in different subspaees, getting the analyses coefficients. Because of the different characteristic between the signals and noises in wavelet packet transform, the analyses coefficients can be processed by the neural network and get a new series of wavelet packet analyses coefficients. Using the new series of coefficients, we can reconstruct a new signal which is the original signals without noises.
出处
《弹箭与制导学报》
CSCD
北大核心
2007年第4期83-85,101,共4页
Journal of Projectiles,Rockets,Missiles and Guidance
关键词
小波包
神经网络
雷达导引头
wavclet packet
neural network
radar seeker
