模糊函数主脊切面极坐标域形态特征提取方法

Morphological feature extraction based on the polar transformation of the slice of ambiguity function main ridge

提取能表现信号本征信息的脉内特征参数是解决当前复杂体制雷达辐射源信号分选难题的有效途径。由于模糊函数能较为细致地反映信号的内在信息,提出一种基于极坐标转换的模糊函数主脊切面形态特征提取方法,该方法首先将模糊函数主脊切面变换到极坐标域以形成封闭的几何图像,然后提取该图像的均面积、致密度和似圆度作为信号分选的特征向量。模糊C均值聚类实验表明,所提特征在大于0 dB条件下分选6类典型雷达信号的成功率保持在95.33%以上,即使在信噪比为-2 dB,分选成功率仍可达到88.5%;在0-20 dB动态信噪比条件下平均分选成功率为99.7%,证实了所提方法的有效性。特征提取耗时实验证明,所提方法提取单个信号特征仅需1.34 s,具有令人满意的时效性。

Extracting the intra-pulse feature parameters is an effective way to sort complex system radar emitter signal which can represent the signal intrinsic information. Because ambiguity function can reflect the internal information of signal meticulously, this work presents a morphological feature extraction method based on polar coordinate conversion of the slice of ambiguity function main ridge. This method converts the slice of ambiguity function main ridge into a closed geometric image by polar transformation. Then, the average area, density and roundness of the image are extracted as the feature vector of signal sorting. The fuzzy C-means clustering experiment shows that the success rate of signal sorting for the six typical radar signals is 95.33% in the fixed SNR environment of 0 dB and the success rate is 88.5% even in-2 dB. Besides, the success rate is 99.7% in the dynamic SNR environment of 0-20 dB. The validity of the proposed method is proved. In addition, the time consuming of feature extraction indicates that the proposed method only needs 1.34 s to extract a single signal feature, which has satisfactory timeliness.