Double-flow SSD SAR image target detection method based on CFAR guidance

According to the double-current SSD SAR image target detection method based on CFAR guidance provided by the invention, an amplitude feature map and a CFAR indication feature map are extracted through a dual-channel feature extraction network composed of an SAR amplitude image feature extraction network Net-A and a CFAR indication binary image feature extraction network Net-C, and different feature maps are fused through a novel ICSAF module. The characteristic of strong scattering of the target in the SAR image is fully utilized, the feature extraction capability of the target detection network is improved, and the target detection performance is enhanced. Meanwhile, a CFAR indication binary image is used as prior guidance information of a target detection network classification loss function, so that a detector more pays attention to learning of difficult negative samples and positive samples; and then the AR-NMS algorithm is used to solve the problems of false alarm and error suppression brought by the pos