Radar Target Recognition Based on Complex HRRP Using Convolutional Neural Network

Owing to the initial phase of the complex high resolution range profile (HRRP) is sensitive, real HRRP is usually used in the radar automatic target recognition (RATR) field, which partially causes losing information of the data. Based on this, this paper proposed complex HRRP target recognition based on improved convolutional neural network (CNN) algorithm. First, the Lenet-5 network in CNN was improved and the recognition performance of the network was ameliorated. Second, utilizing the excellent characteristics of the CNN structure, the real and imaginary parts of the HRRP as the input two channels were input, which makes full use of the phase information of HRRP data. Finally, the superiority of the proposed method is verified by comparison with different methods. The experimental results of measured data show that the complex HRRP based on CNN has good recognition performance.