基于双流金字塔网络的遥感图像显著性目标检测方法

Remote Sensing Image Salient Object Detection Method Based on Dual-Flow Pyramid Network

显著性目标检测在遥感图像处理中起到关键作用,其通常用于对目标区域进行检测和定位.然而,现有的显著性目标检测方法在用于遥感图像显著性检测时存在特征提取不足、对微小目标检测效果差、受背景信息干扰严重等问题.为了解决这些问题,提出了一种基于双流金字塔网络的遥感图像显著性目标检测方法.所提出的方法包括双流金字塔模块和编解码模块,其中双流金字塔模块用于获取遥感图像的细节和语义特征,编解码模块用于进行特征交互并输出显著性检测结果.在公开的遥感显著性检测数据集ORSSD和EORSSD上的实验结果表明所提出方法在检测精度上优于现有的显著性检测方法,证明了所提出方法的有效性和可行性.

Salient object detection plays a key role in remote sensing image processing,which is usually used to detect and locate the target area. However,the existing saliency detection methods for remote sensing images have problems such as insufficient feature extraction,poor detection effect for small targets,and serious interference by background information. In order to solve these problems,a remote sensing image salient object detection method based on dual-flow pyramid network is proposed. The proposed method includes a dual-flow pyramid module and a codec module. The dual-flow pyramid module is used to obtain the details and semantic features of remote sensing images,and the codec module is used for feature interaction and output the saliency detection results. Experimental results on public remote sensing data sets ORSSD and EORSSD show that the proposed method outperforms the existing methods in terms of detection accuracy,which proves the effectiveness and feasibility of the proposed method.