用于雷达辐射源识别的特殊色彩四维度视图

Special color four-dimensional view for radar emitter identification

针对模糊函数在作为雷达辐射源识别深度学习模型输入时存在信息损失严重的问题,提出一种特殊色彩映射四维度视图输入方式。首先将模糊函数映射到红-绿对立色彩空间中,再拼接其三视图与主脊切面以形成识别模型的输入,这样的输入方式更符合模糊函数的分布与深度神经网络处理图片的特点。仿真实验结果表明,将特殊色彩四维度视图作为AlexNet网络的输入时,在SNR=-6dB时准确率仍为88.71%,相较于其他输入方式准确率提高14.61%以上。新的输入方式显著提高了识别模型的准确率和鲁棒性,可作为雷达辐射源智能识别系统的输入方式。

In order to solve the problem of serious information loss when ambiguity function is used as input of radar emitter recognition deep learning model,a special input method of color mapping four-dimensional view is proposed.Firstly,the ambiguity function is mapped to the red-green color space,and then the input of the recognition model is formed by splicing the three views and the main ridge.This input method is more consistent with the distribution of ambiguity function and the characteristics of deep network image processing.The simulation results show that when the special color four-dimensional view is used as the input of AlexNet network,the accuracy is 88.71% when SNR=-6dB,which is 14.61% higher than other input methods.The new input method significantly improves the accuracy and robustness of the identification model and can be used as an input method for the intelligent identification system of radar emitter.