自然场景中的实体标志,如商标、交通标志等,易受拍摄角度、所依附物体形变、尺度变化等影响,导致检测精度降低.为此,提出一种注意力引导的标志检测与识别网络(attention guided logo detection and recognition network,AGLDN),联合优...
详细信息
自然场景中的实体标志,如商标、交通标志等,易受拍摄角度、所依附物体形变、尺度变化等影响,导致检测精度降低.为此,提出一种注意力引导的标志检测与识别网络(attention guided logo detection and recognition network,AGLDN),联合优化模型对多尺度变化和复杂形变的鲁棒性.首先通过标志模板图像搜集及掩码生成、标志背景图像选取和标志图像生成创建标志合成数据集;然后基于RetinaNet和FPN提取多尺度特征并形成高级语义特征映射;最后利用注意力机制引导网络关注标志区域,克服目标变形对特征鲁棒性的影响,实现标志检测与识别.实验结果表明,所提方法可以有效降低尺度变化、非刚性形变的影响,提高标志检测准确率.
为了解决遥感图像变化检测中变化边缘检测不清晰以及极小目标的漏检问题,本论文提出了一种基于注意力引导的多尺度特征融合的遥感图像变化检测网络(MFM-CDNet),旨在提高变化检测的效率和准确性。MFM-CDNet通过特征增强模块FEM和注意力引导的特征融合模块FFM,实现了从粗到细的多尺度特征提取与融合。FEM利用双重注意力机制精确定位图像中的重要区域,而FFM通过对多尺度特征的有效整合与优化,增强了模型对正样本的识别能力,并提高了边缘变化区域的定位精度。在LEVIR-CD、WHU-CD和GoogleGZ-CD三个公开数据集上的实验表明,MFM-CDNet网络综合性能优于现有的变化检测方法。消融实验进一步验证了FEM和FFM模块在提升检测性能中的关键作用。MFM-CDNet的成功应用展示了其在复杂遥感图像变化检测中的鲁棒性和泛化能力,为未来的遥感图像变化检测研究提供了重要的参考。In order to solve the problems of unclear detection of change edges and missed detection of very small targets in remote sensing image change detection, this thesis proposes an attention-guided multi-scale feature fusion-based remote sensing image change detection network (MFM-CDNet), which aims to improve the efficiency and accuracy of the change detection. MFM-CDNet achieves multi-scale feature extraction and fusion from coarse to fine by means of a feature enhancement module, FEM, and an attention-guided feature fusion module, FFM. FFM achieves multi-scale feature extraction and fusion from coarse to fine. FEM utilises a dual attention mechanism to pinpoint important regions in the image, while FFM enhances the model’s ability to identify positive samples and improves the localisation accuracy of edge-change regions through the effective integration and optimisation of multi-scale features. Experiments on three publicly available datasets, LEVIR-CD, WHU-CD and GoogleGZ-CD, show that the comprehensive performance of the MFM-CDNet network outperforms existing change detection methods. The ablation experiments further validate the key role of FEM and FFM modules in enhancing the detection performance. The successful application of MFM-CDNet demonstrates its robustness and generalisation ability in complex remote sensing image change detection, which provides an important reference for future research on remote sensing image change detection.
暂无评论