论文部分内容阅读
本文研究了短时傅时叶变换后再提取奇异值作为特征矢量的特征提取方法，并对提取出的特征利用神经网络进行了识别。从理论上证明了该方法的合理性，说明该方法是一种良好的特征提取方法，通过计算距离可分性测度结果得出该方法提取的特征比一般傅里叶变换法提取的特征具有更好的可分性。利用此方法提取出的目标特征及傅里叶变换后再提取奇异值的特征抽取方法可以获得更好的识别结果。
In this paper, we study the feature extraction method of short-time Fourier transform and then extract the singular value as feature vector, and identify the extracted features by using neural network. The rationality of this method is proved theoretically, which shows that this method is a good feature extraction method. By calculating the result of distance separability, it can be concluded that the feature extracted by this method has more features than those obtained by general Fourier transform Good separability. Using this method to extract the target features and Fourier transform and then extract the singular value of the feature extraction method can get better recognition results.