Self-supervised remote sensing image space-spectrum fusion method and device based on prior image constraint
By using a priori images as constraints in the null spectral fusion of remote sensing images and using a self-supervised training method, the problem of insufficient consideration of the relationship between full-color images and multi-spectral images in the prior art is solved, and the stability and accuracy of the fusion results are improved.
Patent Information
- Application Number
- CN202411782510.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-04-18
- Estimated Expiration
- 2044-12-05
AI Technical Summary
The existing deep learning methods lack sufficient consideration of the inherent relationship between full-color and multi-spectral images when fusion of remote sensing images, resulting in distortion of the fusion results in spatial details and spectral information maintenance, especially when processing different sensor data, the generalization ability of the model is poor.
The self-supervised remote sensing image null spectrum fusion method based on prior image constraints is adopted. By generating a prior image, the image contains the spatial information of the full-color image and the spectral information of the multi-spectral image, and is input into the remote-sensing image null spectrum fusion network as constraints. The self-supervised training method is used to calculate the probability distribution differences between the spatial loss and spectral loss between the fusion image, the sample full-color image and the sample multi-spectral image through the target loss function.
The generalization ability of the model when processing different sensor data is improved, a more stable fusion effect is achieved, and the accuracy of fusion results in spatial details and spectral information is enhanced.