基于时频图的飞机目标特征提取算法

Feature-extraction Algorithm of Airplane Targets Based on Time-Frequency Diagram

研究了两种基于时频分析的飞机目标特征提取算法。首先,介绍利用短时傅里叶变换获得目标二维时频图信息,然后,分析飞机目标回波时频图的图像熵特征和平均时频谱特征,给出了这两种特征的提取算法,仿真得到利用这些特征对三类飞机的分类识别结果,分析了其目标分类识别性能以及适用情况。在研究时频图平均时频谱特征提取算法时,发现单独使用此特征在飞机目标分类识别中存在的问题,加入基于平均时频谱的方差特征,综合两种特征实现了对三类飞机目标的的分类识别。仿真结果表明,这两种特征的目标识别正确率都能达到94%以上。

Two target feature extraction of airplane is mainly studied by time-frequency diagram（TFD}. The first part based on air- plane backscatter TFD I which comes from STFT, two algorithms for airplane are proposed and studied, one of them using target image entropy characteristics, anther using average time-frequency entropy and variance characteristics. As the same time, the merits and drawbacks of this method has been simulated and explained. This two feature have effective detection performance, the correctly rate all above 94 percent.