Long-range drone image sub-scene segmentation method, system, medium and equipment
By constructing and preprocessing the initial scene map, the image camera parameters are obtained to optimize the gradient direction, and the connection edge weight is set using the normalized graph cutting algorithm, the robustness problem of sub-scene segmentation in the long-range drone image is solved, and high-precision sub-scene reconstruction is achieved.
Patent Information
- Application Number
- CN202510487247.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-18
- Publication Date
- 2025-08-22
- Estimated Expiration
- 2045-04-18
AI Technical Summary
In the prior art, in the long-range belt drone images, the sub-scene segmentation method cannot effectively reflect the connection point error, resulting in low sub-scene reconstruction accuracy. The traditional method cannot guarantee the robustness of sub-scene segmentation when the long-range belt image air belt constraint is weak.
By constructing the initial scene map and preprocessing, the parameters of a single image camera are obtained to optimize the gradient direction, and the weight information of the image connected edges and the normalized graph cutting algorithm are used to segment the subscene to ensure the accuracy of the weight of the connected edges and the robustness of the segmentation.
The robustness and accuracy of neutron scene segmentation of long-distance drone images is improved, providing a stable internal structure for subsequent sub-scene reconstruction, and supporting high-precision and rapid reconstruction.