基于模糊平面的信号识别方法

将一维信号变换到二维坐标平面往往更有利于描述信号的时变特征,从而实现信号的分类识别。基于离散时频分布的信号识别方法,将时频核设计问题转化为以信号自模糊函数为原始特征的特征选择问题,以实现特征降维和信号识别。时频核设计孤立考察模糊平面上各个特征点,且降维空间中存在着识别信息冗余。将核设计的原理推广,直接基于模糊平面进行信号识别,利用K—L展开和线性变换对自模糊函数进行特征提取,在降维空间内综合了各原始特征共有的分类信息,并去除特征之间的相关性,从而比时频核设计方法具有更优的信号识别性能。

Signal Recognition Method Based on Ambiguity Plane

The time - varying properties can be extracted better when signal being transformed on 2D coordinate plane for the purpose of signal recognition and classification. Actually the kernel design in the recognition method based on discrete time - frequency representation is a problem of feature selection from the ambiguity functions to reduce feature dimension. Each point on ambiguity plane is considered independently when kernel designing. The feature space of reduced dimension contains the classification information redundancy. The principle of kernel design is generalized in the paper and signals are recognized on the ambiguity plane directly. With the feature extraction using K - L expansion and linear transformation with the auto-ambiguity functions, the common recognition information is integrated in the new feature space. Moreover, the correlations between features elements are eliminated, so recognition performances are improved compared with the kernel design method.