基于谱-空跨维度注意力机制的高光谱图像波段选择方法及系统
By employing a hyperspectral image band selection method based on a spectral-spatial cross-dimensional attention mechanism, and utilizing spectral-spatial cross-dimensional correlation information, the method addresses the problem of unreasonable band selection in existing methods, achieving more efficient band selection and improved performance of downstream tasks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QINGDAO UNIV
- Filing Date
- 2024-01-11
- Publication Date
- 2026-07-17
AI Technical Summary
Existing hyperspectral image band selection methods ignore the interrelationship between spectral and spatial dimensions, resulting in unreasonable band selection results, wasted computational resources, high data redundancy, and impact on the performance of downstream tasks.
A hyperspectral image band selection method based on the spectral-spatial cross-dimensional attention mechanism is adopted. By using a band attention module, a spectral-spatial cross-dimensional attention module, and an autoencoder, the method fully utilizes the spectral-spatial cross-dimensional correlation information to generate a band attention-weighted image and optimize the model.
This improved the rationality and quality of band selection results, enhanced the performance of downstream tasks, and reduced computational resource consumption and data redundancy.
Smart Images
Figure CN117853920B_ABST