基于Faster-RCNN的IR-UWB穿墙雷达邻近多目标检测算法

Multiple Adjacent Targets Detection Algorithm for IR-UWB Through-wall Radar Based on Faster-RCNN

针对超宽带冲激脉冲(Ultra-wideband Impulse Radio, IR-UWB)穿墙雷达邻近多目标精确检测难点,提出了一种基于Faster-RCNN的IR-UWB穿墙雷达邻近多目标检测算法。算法采用自适应预处理算法对雷达回波进行处理,凸显疑似目标图像域特征;采用Faster-RCNN网络对不同目标距离间隔下的邻近多目标进行分离检测;利用穿墙雷达实测数据与传统的恒虚警率(Constant False-Alarm Rate, CFAR)检测方法进行对比试验,结果表明该算法明显优于传统方法。

To overcome the difficulty of accurate detection of multiple adjacent targets by Ultra-wideband Impulse Radio(IR-UWB) through-wall radar, a multiple adjacent targets detection algorithm for IR-UWB through-wall radar based on Faster-RCNN is proposed. Firstly, the adaptive preprocessing algorithm is used to process the radar echo and highlight the image domain features of the suspected target. Secondly, the Faster-RCNN network is used to detect and separate multiple adjacent targets at different target distances. Finally, the algorithm is compared with the traditional Constant False-Alarm Rate(CFAR) detection method by using the measured data of the through-wall radar. The results show that the proposed algorithm is significantly better than the traditional method.