一种图像配准方法、装置、电子设备及存储介质
By performing super-resolution reconstruction and homography estimation on cross-modal and cross-resolution images, the registration problem of multimodal images under modal and resolution differences is solved, and high-precision image alignment is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIHANG UNIV
- Filing Date
- 2024-01-31
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot effectively solve the problem of homography estimation for image registration when there are differences in modality and resolution in multimodal images, resulting in low registration accuracy.
By acquiring original and reference images with different modalities and resolutions, downsampling is performed to generate intermediate images, and super-resolution reconstruction is performed based on the intermediate images. The homography matrix between the super-resolution image and the intermediate image is calculated, and the homography matrix is used to perform distortion operations on the super-resolution image to achieve cross-modal and cross-resolution image registration.
High-precision registration of multimodal images with different resolutions is achieved. The resolution gap between images is reduced through super-resolution reconstruction, and the homography matrix is predicted by multimodal homography estimation to achieve image alignment.
Smart Images
Figure CN118071803B_ABST