论文部分内容阅读
本文采用小波变换把毫米波辐射计的目标信号分解为一系列正交子空间上的投影，利用重构矩法分析信号特征在各子空间的分布，依次提取各子空间上的特征，然后融合这些特征，组成特征矢量，采用神经网络对目标信号特征矢量进行建模．用此方法对低信噪比的毫米波辐射计的信号进行仿真试验，结果表明该方法克服了传统方法对噪声和目标信号散布的敏感，取消了对目标和辐射计天线之间距离的限制，与最近邻法相比，该方法大大提高了识别率．
In this paper, the wavelet transform is used to decompose the target signal of a millimeter wave radiometer into a series of projections on an orthogonal subspace. The reconstruction moments are used to analyze the distribution of the signal features in each subspace, and then the features in each subspace are sequentially extracted and then fused These features make up the feature vector and use neural network to model the target signal feature vector. This method is used to simulate the signal of millimeter wave radiometer with low signal to noise ratio. The results show that this method overcomes the sensitivity of the traditional method to the noise and the spread of the target signal, cancels the limit of the distance between the target and the radiometer antenna, Compared with the nearest neighbor method, this method greatly improves the recognition rate.