基于空域和频域特征融合的轻量化SAR图像目标检测OA

To address the limitations of current Synthetic Aperture Radar(SAR)image target detection methods,such as the single mode of feature extraction and susceptibility to speckle noiseinterference,a lightweight detection model Spatial-Frequency Converse2D GroupHead YOLO(SFCG-YOLO)is proposed based on YOLOv11,which integrates spatial and frequency domain features and enhances edge information.This method designs a spatial-frequency feature fusion module,combining spatial convolution to extract spatial information and frequency convolution to capture frequency information,thereby enhancing feature representation capabilities.It introduces the depthwise separable inverse convolution operator Converse2D to reduce upsampling feature loss,suppress noise,and restore target edge details.A lightweight detection head is constructed,redesigned using grouped convolution to reduce the number of model parameters and computational overhead.Experimental results on the HRSID dataset demonstrate that the proposed method achieves an mean Average Precision(mAP)of 92.9%(mAP50)and 69.92%(mAP50-95).Compared with the original YOLOv11,the proposed model has lower complexity and achieves a balance between detection accuracy and model efficiency,making it suitable for SAR target detection tasks in complex environments.