基于特征引导的遥感图像显著性目标检测  

Feature guided remote sensing image saliency target detection

作　　者：顾军华 崔彭滔[1] 徐雯佳 Gu Junhua;Cui Pengtao;Xu Wenjia(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300401,China;Hebei Key Laboratory of Big Data Calculation(Hebei University of Technology),Tianjin 300401,China;Hebei Institute of Hydrogeology&Engineering Geological Survey,Shijiazhuang 050021,China)

机构地区：[1]河北工业大学人工智能与数据科学学院,天津300401 [2]河北省大数据计算重点实验室(河北工业大学),天津300401 [3]河北省水文工程地质勘查院(河北省遥感中心),石家庄050021

出　　处：《现代计算机》2023年第20期9-15,共7页Modern Computer

基　　金：河北省自然资源厅计划项目(454-0601-YBN-IBBM)。

摘　　要：基于遥感图像的显著性目标检测在建筑物、飞机、船舶、河流、岛屿等检测任务中取得了显著成效。遥感图像尺度多变、场景复杂,在显著性目标检测过程中,发现一些遥感图像显著性检测算法会出现显著目标误判问题,即网络模型容易将显著目标的部分区域判断为背景,通过引入亚像素卷积解码器和特征引导解码器,在F度量、S度量等指标上提升幅度分别为2.83个百分点、3.48个百分点,平均绝对误差下降了0.5个百分点,证明了模型的有效性。Saliency object detection based on remote sensing images has achieved remarkable results in the detection tasks of buildings,aircraft,ships,rivers,islands and so on.In the process of saliency target detection,it is found that some remote sensing image saliency detection algorithms will have the problem of misjudgment of significant targets,that is,the network model is easy to judge part of the salient target as the background,and by introducing subpixel convolutional decoder and feature guidance decoder,the improvement range of F metric and S-measure is 2.83 percentage points and 3.48 percentage points,respectively,and the average absolute error decreases by 0.5 percentage points,which proves the effectiveness of the model.

关 键 词：遥感图像 深度学习 显著性检测 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]