In order to improve the quality of radar scattering image, image pre- processing methods such as enhanced filtering, contrast Adaptive histogram equalization and energy normalization are adopted to meet the needs of precise detection and recognition of vehicle targets in the ground clutter environment Two parameters are innovatively constructed and an improved YOLO target recognition method is constructed by using non-maximum suppression method, based on the airborne radar scattering image, a vehicle target radar automatic recognition verification test is carried out. The results show that the improved YOLO algorithm can improve the target recognition rate by 12% compared with traditional algorithms such as deformation convolutional neural network (DPM), and has the ability to identify weakly scattering targets, it provides a way to evaluate the performance of radar target feature control.