基于深度注意力网络和场景感知的遥感图像压缩方法

By employing a deep attention network and a scene-aware remote sensing image compression algorithm, and utilizing channel-spatial attention residual blocks and an adaptive scene-aware framework, the problems of scene differences and non-local self-similarity in remote sensing image compression are solved, achieving high-quality remote sensing image reconstruction and bitrate optimization.

CN115511983BActive Publication Date: 2026-07-17SICHUAN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SICHUAN UNIV
Filing Date
2021-06-07
Publication Date
2026-07-17

Smart Images

  • Figure CN115511983B_ABST
    Figure CN115511983B_ABST
Patent Text Reader

Abstract

本发明公开了一种基于深度注意力网络和场景感知的遥感图像压缩算法，其实现过程包括：将编码和解码过程建模为卷积神经网络，让经过预处理的原始图像通过含有通道‑空间注意力残差组的图像编解码网络；构建自适应场景感知框架：由构建的自动编解码器训练得到初始模型m，依次用不同场景数据库训练得到不同场景类型的类别模型测试阶段，对待处理的遥感图像进行场景匹配后进行图像压缩。实验结果表明，引入注意力特征提取模块能有效提升性能，进一步的，采用场景感知策略可以增强对特定场景类别的压缩处理效果。所提出的框架能有效地提升编码图像的率失真性能，并能获得更好的视觉效果。
Need to check novelty before this filing date? Find Prior Art