图像特征优化融合方法、全色锐化方法及产品
By using an image feature optimization fusion method based on adaptive state space and wavelet linear attention, the contradiction between spectral fidelity and spatial detail restoration in traditional remote sensing image fusion is resolved, generating a high-resolution multispectral image that balances spectral consistency and spatial detail.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEFEI UNIV OF TECH
- Filing Date
- 2026-03-18
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional remote sensing image fusion technology presents a contradiction between spectral fidelity and spatial detail restoration. It is difficult to balance the excessive suppression of high-frequency information and the insufficient smoothness of low-frequency structures, resulting in a fusion result that fails to achieve both spectral fidelity and spatial detail restoration.
An image feature optimization fusion method based on adaptive state space and wavelet linear attention is adopted. The low-frequency structure and high-frequency details of PAN and LRMS features are enhanced by HCB processing, and deep fusion of cross-modal features is achieved by fusion block. Combined with the reconstruction head, a high-resolution multispectral image is generated.
It achieves a balance between spectral fidelity and spatial detail restoration in remote sensing image fusion, and the generated high-resolution multispectral images exhibit excellent performance in spectral consistency and spatial detail, making them suitable for various satellite data.
Smart Images
Figure CN121904538B_ABST