利用改进特征金字塔模型的SAR图像多目标船舶检测

深度学习模型中的特征金字塔网络（Feature Pyramid Network，FPN）常被用作合成孔径雷达（Synthetic Aperture Radar，SAR）图像中多目标船舶的检测。针对复杂场景下多目标船舶检测问题，提出了一种基于改进锚点框的FPN模型。首先将特征金字塔模型嵌入传统的RPN（Region Proposal Network）并映射成新的特征空间用于目标检测，然后利用基于形状相似度距离（Shape Similar Distance,SSD）度量的Kmeans聚类算法优化FPN的初始锚点框，并使用SAR船舶数据集测试。实验结果表明，所提算法目标检测精确率达到98.62%，在复杂场景下与YOLO、Faster RCNN、FPN based on VGG/ResNet等模型进行对比，模型准确率提高，整体性能更好。

Multi-target Ship Detection of SAR Images by Using Improved Feature Pyramid Networks Model

Feature pyramid network(FPN) in deep learning models is commonly used for detection of multiple small ship objects in synthetic aperture radar(SAR) images.In order to solve the problem of small-object and multi-object ship detection under complex scenarios,a ship object detection method based on an improved FPN model is proposed.The feature pyramid model is first embedded into a traditional region proposal network(RPN) and mapped into a new feature space for object detection.Then,the K-means clustering algorithm based on the shape similar distance(SSD) measure is used to optimize the FPN initial anchor box and perform test by using SAR ship dataset.Experimental results show that the object detection accuracy of the proposed algorithm reaches 98.62%.Compared with YOLO,RPN based on VGG/ResNet,FPN based on VGG/ResNet and other models in complex scenarios,the model proposed shows higher accuracy and better overall performance.